Automatic adjustment of data access policies in data analytics

Abstract

An autoencoder network is generated from the first model parameters. A reconstruction error for the autoencoder network is measured, the reconstruction error comprising a difference between an input to the autoencoder network and a corresponding output from the autoencoder network, the input to the autoencoder network comprising a portion of an initial set of data. The reconstruction error and a confidence score corresponding to a complexity level of the autoencoder network are aggregated into a level of difficulty score for the autoencoder network. A derived data access policy level corresponding to the initial data access policy level is generated from the level of difficulty score and an initial data access policy level corresponding to the initial set of data, where the derived data access policy level provides access to a transformed data set generated by applying a transformation to the initial set of data.

Description

[Technical field]

[0001] The present invention relates generally to a method, system, and computer program product for adjusting data access policies, and more particularly, to a method, system, and computer program product for automatically adjusting data access policies in data analytics. [Background technology]

[0002] In a data analytics pipeline, input data is received and processed, and results are output. Some implementations include multiple processing steps, or layers, where the output of one step is used as the input for the next step in the pipeline or in a series of sequences. Each processing step is called a transformation, but the data itself does not need to be changed. In some implementations, such as when processing time series data, the input data is a continuous stream and the output data is the result of processing the input stream. In other implementations, the input data is not a continuous stream but is discrete batches of data, and the output data is also discrete batches of data.

[0003] In data analytics, a column (also called an attribute or data attribute) is a set of data values ​​of a particular type. A row holds the data for each instance of a column. Thus, the intersection of a row and a column holds a particular data value for each attribute. For example, a set of employee data might have columns labeled Name, Salary, Month of Birth, Address, and Phone. Each row of data might hold data for one particular employee, such as Jane Smith's name, salary data, month of birth (for use in planning team building events), address, and mobile phone number. Data attributes are also referred to as columns, even if the data is not actually stored in a two-dimensional row-and-column format.

[0004] Data access policies also control which entities can access specific data. In one data access policy implementation, each data column or attribute is assigned a data access level on a scale from least protected (e.g., 1) to most protected (e.g., 5). Data accessing entities are also assigned data access levels on the same scale depending on a combination of their ability to protect and their need to know the protected data. Only data accessing entities with a data access level equal to or higher than the data access level of the attribute are allowed to access that data. For example, a data access system assigned to an administrative assistant may be assigned a data access level of 1 and be allowed to access data also assigned a data access level of 1, such as employee birth month data, but not data assigned a higher data access level. However, a payroll processing system may be assigned a data access level of 4 and may be allowed to access data assigned data access levels 1-4, such as the data listed above including employee payroll data (which has data access level 4). Summary of the Invention [Means for solving the problem]

[0005] Exemplary embodiments provide methods, systems, and computer program products. One embodiment includes a method of generating an autoencoder network from a first model parameter. One embodiment measures a reconstruction error for the autoencoder network, where the reconstruction error includes a difference between an input to the autoencoder network and a corresponding output from the autoencoder network, where the input to the autoencoder network includes a portion of an initial set of data. One embodiment aggregates the reconstruction error and a confidence score corresponding to a complexity level of the autoencoder network into a level of difficulty score for the autoencoder network. One embodiment generates a derived data access policy level corresponding to the initial data access policy level from the level of difficulty score and an initial data access policy level corresponding to the initial set of data, where the derived data access policy level implements access to a transformed data set generated by applying a transformation to the initial set of data.

[0006] An embodiment includes one or more computer-usable program products comprising one or more computer-readable storage devices and a plurality of program instructions stored on at least one of the one or more computer-readable storage devices.

[0007] An embodiment includes a computer system having one or more processors, one or more computer readable memories, and one or more computer readable storage devices, and a plurality of program instructions stored on at least one of the one or more computer readable storage devices for execution by at least one of the one or more processors via at least one of the one or more computer readable memories.

[0008] Certain novel features which are believed to be characteristic of the invention are set forth in the appended claims, however the invention itself, together with its preferred modes of use, further objects and advantages thereof, will best be understood by reference to the following detailed description of illustrative embodiments when read in connection with the accompanying drawings. [Brief description of the drawings]

[0009] [Figure 1] FIG. 1 illustrates a block diagram of a network of data processing systems in which illustrative embodiments may be implemented. [Diagram 2] FIG. 2 illustrates a block diagram of a data processing system in which illustrative embodiments may be implemented. [Diagram 3] FIG. 3 illustrates a block diagram of an example configuration for automatically adjusting data access policies in data analytics, according to an example embodiment. [Figure 4] FIG. 4 illustrates a block diagram of an example configuration for automatically adjusting data access policies in data analytics, according to an example embodiment. [Diagram 5] FIG. 5 illustrates an example of automatically adjusting data access policies in data analytics, according to an exemplary embodiment. [Figure 6] FIG. 6 illustrates a continuing example of automatically adjusting data access policies in data analytics, according to an exemplary embodiment. [Figure 7] FIG. 7 illustrates a continuing example of automatically adjusting data access policies in data analytics, according to an exemplary embodiment. [Figure 8] FIG. 8 illustrates a continuing example of automatically adjusting data access policies in data analytics, according to an exemplary embodiment. [Figure 9] FIG. 9 illustrates a continuing example of automatically adjusting data access policies in data analytics, according to an exemplary embodiment. [Figure 10]FIG. 10 illustrates a continuing example of automatically adjusting data access policies in data analytics, according to an exemplary embodiment. [Figure 11] FIG. 11 illustrates a flowchart of an example process for automatically adjusting data access policies in data analytics, according to an example embodiment. [Figure 12] FIG. 12 illustrates a cloud computing environment according to one embodiment of the invention. [Figure 13] FIG. 13 illustrates abstraction model layers according to one embodiment of the present invention. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0010] The exemplary embodiments recognize that if the input data to an analytics operation is protected by a data access policy, then the output of the analytics operation (also called derived data) may also require appropriate protection. If the derived data is not adequately protected, unauthorized entities may access and analyze the derived data to recover the original input data. However, protecting data that is already public, or that poses no risk if public, is unnecessary and complicates data processing operations. Thus, the exemplary embodiments recognize that there is a need to determine a data access policy for derived data.

[0011] Exemplary embodiments also recognize that currently, human experts enforce data access policies on derived data, including, for example, manually setting a particular data access policy level, based on a combination of simple categorical rules (e.g., all health-related data or all salary-related data must have at least a particular data access policy level) and expert intuition, regardless of whether the person believes the derived data is sufficiently anonymized. However, expert judgments can change over time, and different experts often produce inconsistent results. Also, analytics operations are often many-to-many rather than one-to-one, and become more complex over time as operations are added to the analytics pipeline. As a result, data access policy analysis becomes too complex for humans to perform accurately, resulting in overprotection or underprotection. In addition, expert judgments rarely take into account the particular difficulty of analyzing a particular set of derived data to recover the original data. As a result, the illustrative embodiments recognize that there is a need to automatically determine the data access policy level of the derived data based on the data access policy level of the original data and the difficulty of recovering the original data from the derived data.

[0012] The illustrative embodiments recognize that currently available tools or solutions do not address these needs or provide adequate solutions to these needs. The illustrative embodiments used to describe the present invention generally address and solve the above-mentioned problems and other problems associated with automatically adjusting data access policies in data analytics.

[0013] Embodiments can be implemented as a software application. An application implementing an embodiment can be configured as a modification to an existing data analytics system, as a separate application that operates in conjunction with an existing data analytics system, as a standalone application, or some combination thereof.

[0014] In particular, some illustrative embodiments provide a method for determining a level of a difficulty score associated with recovering the original data from the derived data, and generating a derived data access policy level from the difficulty score and an initial data access policy level corresponding to the initial set of data.

[0015] One embodiment measures a level of difficulty score associated with recovering original data from derived data by determining an inverse transformation of a transformation from columns or attributes of original data to columns or attributes of derived data. In particular, the embodiment uses a set of original data and a set of data derived from the original data to fit a model to the inverse transformation, i.e., the transformation from derived data to original data. One embodiment fits the model by performing a predetermined series of regression steps. At the end of the regression steps, the embodiment determines an error metric of the inverse transformation. One non-limiting example of an error metric is the difference between the model's prediction of the original data and the actual original data. One embodiment converts the error metric into a level of difficulty score. The level of difficulty score represents the difficulty of restoring the original data from the derived data. The lowest level of difficulty score within a predetermined range represents derived data that is equal to the original data or original data that is easy to restore from the derived data. A higher level of difficulty score represents a correspondingly more difficult recovery operation, due to the processing time required to generate a sufficiently accurate model or the complexity of the generated model. One embodiment converts the error metric to a level of difficulty score by normalizing it from zero to one, where zero represents the lowest level of difficulty score and one represents the highest level of difficulty score. Other techniques for converting the error metric to a level of difficulty score, and other difficulty score ranges, are also possible and contemplated within the scope of exemplary embodiments.

[0016] However, measuring the level of the difficulty score associated with recovering the original data from the derived data by determining the difficulty of converting the derived data back to the original data is only possible if there is a one-to-one relationship between the original data and the derived data. Consequently, alternative embodiments use alternative methods to measure the level of the difficulty score.

[0017] An autoencoder network is a lossy data compression algorithm that includes an encoder portion and a decoder portion. The encoder portion converts a unit of input data (e.g., the intersection of a row and a column) into an encoded form, and the decoder portion converts the encoded form into a reconstructed data. The difference between the reconstructed data and the original input data is called the reconstruction loss or reconstruction error. One or more model parameters define a particular configuration of the autoencoder network. For example, if both the encoder and decoder portions are implemented as fully connected neural networks with one or more layers, some non-limiting examples of model parameters are the number and size of hidden layers in each of the encoder and decoder portions, the number of dimensions of the encoded form of the data, the number of training epochs used to train the specially constructed autoencoder network, the loss function used to train the specially constructed autoencoder network, the rate at which the specially constructed autoencoder network should train, the dropout rate (the probability of learning a given node in a layer), and any learning rate optimizer used during training. The loss function measures the difference between the actual output of the neural network and the desired output of the neural network, and thus seeks to minimize the loss function by training the neural network.

[0018] In one embodiment, the autoencoder network is generated from one or more model parameters, which in one embodiment are set to default values, set by a human expert based on knowledge of the data to be processed by the analytics operation, automatically generated, or set using another currently known technique, or set using a combination of these techniques.

[0019] One embodiment trains the generated autoencoder network according to one or more model parameters. One embodiment uses a set of original data and a set of data derived from the original data as training data. One embodiment uses the reconstruction error of the autoencoder network when processing a training set of original data as a loss function during training. Another embodiment uses the difference between an encoded form of the original training data (the output of an encoder part of the autoencoder network) and the corresponding derived training data as a loss function during training. Another embodiment trains the autoencoder network to minimize both loss functions.

[0020] One embodiment measures the reconstruction error and complexity level of the generated autoencoder network once it has trained the generated autoencoder network for a specified number of training epochs. To measure the complexity level of a network, one embodiment combines model parameters related to the complexity of the model into a common complexity level. One embodiment adds to the complexity level the number and size of hidden layers in each of the encoder and decoder portions and the number of dimensions of the encoded form of the data. Another embodiment also includes in the complexity level measurement the number of training epochs used in training a specially constructed autoencoder network.

[0021] Because a more complex autoencoder network should have a smaller reconstruction error than a less complex autoencoder network when processing the same data, one embodiment uses the complexity level as a confidence score that corresponds to the reconstruction error of the generated autoencoder network.

[0022] Another embodiment generates a set of autoencoder networks by varying one or more model parameters. For example, one embodiment may generate three autoencoder networks, each with a different number of hidden layers, a different dimensionality of the data encoded form, or both. One embodiment trains the generated set of autoencoder networks according to each network's model parameters, and measures the reconstruction error and complexity level of each autoencoder network in the manner described herein. By analyzing multiple autoencoder networks, it is often possible to more accurately determine the level of difficulty of recovering the initial data using derived data output from the analytics operation, at the expense of additional analysis time.

[0023] One embodiment aggregates the reconstruction errors and corresponding complexity levels of one or more autoencoder networks into a level of difficulty score. One embodiment calculates the level of difficulty score by averaging the reconstruction errors of one or more autoencoder networks. Another embodiment calculates the level of difficulty score by averaging the reconstruction errors of one or more autoencoder networks, where each autoencoder network has a complexity level above or below a complexity level threshold. Another embodiment uses the reconstruction error of the autoencoder network with the lowest reconstruction error as the level of difficulty score. Another embodiment uses the reconstruction error of the autoencoder network with the lowest reconstruction error as the level of difficulty score, where the autoencoder network also has a complexity level above or below a complexity level threshold. The level of difficulty score represents a level of difficulty associated with recovering the initial data using derived data output from an analytics operation.

[0024] One embodiment determines a level of difficulty score for autoencoder networks generated from an initial set of model parameters, varies one or more model parameters from their initial settings, and generates a corresponding set of autoencoder networks. One embodiment trains the generated set of autoencoder networks according to each network's model parameters, measures the reconstruction error and complexity level of each network, and aggregates the reconstruction error and corresponding complexity level of the autoencoder networks into a new level of difficulty score in a manner described herein. Determining an initial level of difficulty score as a baseline and generating and analyzing additional networks based on that baseline helps limit the generation of additional networks to those predicted to be most useful in assessing the overall level of difficulty.

[0025] One embodiment uses the level of the difficulty score and the data access policy level of the initial data to determine the data access policy level of the derived data. For example, if the level of the difficulty score is low (e.g., 1 on a scale of 0 to 5), it indicates that it is not difficult at all to recover the initial data from the derived data, and the data access policy level of the derived data should be equal to the data access policy level of the initial data. On the other hand, if the level of the difficulty score is high (e.g., 4.5 on a scale of 0 to 5), it indicates that it is very difficult to recover the initial data from the derived data, and the data access policy level of the derived data can be lowered, i.e., relaxed, relative to the data access policy level of the initial data.

[0026] The manner of automatically adjusting data access policies in data analytics described herein is not available in currently available methods in the technical field related to data analytics. The method of the embodiments described herein, when implemented to run on a device or data processing system, comprises a substantial advance in the capabilities of the device or data processing system in generating an autoencoder network from model parameters, measuring a reconstruction error of the autoencoder network, aggregating the reconstruction error and confidence scores into a level of difficulty score for the autoencoder network, and generating a data access policy level derived from the level of difficulty score and an initial data access policy level corresponding to the initial set of data.

[0027] Exemplary implementations are described by way of example only with respect to certain content, transmission, delay, events, climax events, non-climax events, durations, predictions, thresholds, validation, responses, rankings, adjustments, sensors, measurements, devices, data processing systems, environments, components, and applications. Any specific representation of these and other similar artifacts is not intended to limit the invention. Any suitable representation of these and other similar artifacts may be selected within the scope of exemplary implementations.

[0028] Moreover, the exemplary embodiments may be implemented with respect to any type of data, data source, or access to a data source on a data network. Any type of data storage device may provide data to an embodiment of the invention locally at a data processing system or over a data network within the scope of the present invention. Where an embodiment is described using a mobile device, any type of data storage device suitable for use with a mobile device may provide data to such an embodiment within the scope of the exemplary embodiments, either locally at the mobile device or over a data network.

[0029] The exemplary embodiments are described by way of example only, using specific code, designs, architectures, protocols, layouts, schematics, and tools, without limiting the exemplary embodiments. Moreover, the exemplary embodiments may be described by way of example only, using specific software, tools, and data processing environments for clarity of explanation. The exemplary embodiments may be used in combination with other equivalent or similar purpose structures, systems, applications, or architectures. For example, other equivalent mobile devices, structures, systems, applications, or architectures therefor may be used in combination with such embodiments of the present invention within the scope of the present invention. The exemplary embodiments may be implemented in hardware, software, or a combination thereof.

[0030] The examples in this disclosure are used for clarity of explanation only and are not intended to be limiting of the exemplary embodiments. Additional data, operations, actions, tasks, activities, and operations are contemplated by this disclosure, and the same are contemplated within the scope of the exemplary embodiments.

[0031] Any advantages listed herein are exemplary only and are not intended to be limiting to example embodiments. Additional or different advantages may be realized by certain example embodiments. Moreover, certain example embodiments may have some, all, or none of the advantages listed above.

[0032] Although the present disclosure includes detailed descriptions of cloud computing, it should be understood in advance that implementation of the teachings recited herein is not limited to a cloud computing environment. Rather, embodiments of the present invention can be implemented in conjunction with any other type of computing environment now known or later developed.

[0033] Cloud computing is a model of service delivery for enabling convenient, on-demand network access to a shared pool of configurable computing resources (e.g., networks, network bandwidth, servers, processing, memory, storage, applications, virtual machines, and services) that can be rapidly provisioned and released with minimal administrative effort or interaction with the service provider. The cloud model may include at least five characteristics, at least three service models, and at least four deployment models.

[0034] The features are as follows.

[0035] On-demand self-service: A cloud consumer can unilaterally provision computing capacity, such as server time and network storage, automatically as needed, without requiring human interaction with the provider of the service.

[0036] Broad Network Access: Capabilities are available over the network and accessed via standard mechanisms that facilitate use by heterogeneous thin or thick client platforms (e.g., cell phones, laptops, and PDAs).

[0037] Resource Pooling: Provider computing resources are pooled to serve multiple consumers using a multi-tenant model, and various physical and virtual resources are dynamically allocated and reallocated according to demand. Consumers generally have no control or knowledge of the exact location of the provided resources, but are said to be location-independent in that they may be able to specify a location at a higher level of abstraction (e.g., country, state, or data center).

[0038] Rapid Elasticity: Capabilities can be provisioned quickly and elastically, sometimes automatically, quickly scaled out, quickly released and quickly scaled in. To the consumer, the capabilities available for provisioning are often unlimited and can be purchased in any quantity at any time.

[0039] Measured Services: Cloud systems automatically control and optimize resource usage by using metering capabilities at some level of abstraction appropriate to the type of service (e.g., storage, processing, bandwidth, and active user accounts). Resource usage can be monitored, controlled, and reported, providing transparency to both providers and consumers of the services being utilized.

[0040] The service model is as follows:

[0041] Software as a Service (SaaS): The capability offered to consumers to use a provider's applications running on a cloud infrastructure. The applications are accessible from a variety of client devices through a thin-client interface, e.g., a web browser (e.g., web-based email). The consumer does not manage or control the underlying cloud infrastructure, including networks, servers, operating systems, storage, or even individual application functions, with the possible exception of limited user-specific application configuration settings.

[0042] Platform as a Service (PaaS): A capability offered to a consumer to deploy consumer-created or acquired applications, created using programming languages ​​and tools supported by the provider, onto a cloud infrastructure. The consumer does not manage or control the underlying cloud infrastructure, e.g., including networks, servers, operating systems, or storage, but does have control over the deployed applications and, in some cases, the application-hosting environment configuration.

[0043] Infrastructure as a Service (IaaS): The capability offered to consumers to provision processing, storage, network, and other basic computing resources on which the consumer can deploy and run any software, which may include operating systems and applications. The consumer does not manage or control the underlying cloud infrastructure, but does have control over the operating systems, storage, deployed applications, and in some cases, limited control over selecting network components (e.g., host firewalls).

[0044] The deployment models are as follows:

[0045] Private Cloud: Cloud infrastructure is operated exclusively for an organization. The cloud infrastructure may be managed by the organization or a third party, and may be on-premises or off-premises.

[0046] Community Cloud: Cloud infrastructure is shared by several organizations and supports a particular community with common concerns (e.g., mission, security requirements, policies, and compliance considerations). The cloud infrastructure may be managed by the organizations or a third party, and may reside on-premise or off-premise.

[0047] Public Cloud: The cloud infrastructure is available to the general public or large industry groups and is owned by an organization that sells cloud services.

[0048] Hybrid Cloud: The cloud infrastructure remains a unique entity, but is a hybrid of two or more clouds (private, community, or public) brought together by standardized or specialized technologies that enable data and application portability (e.g., cloud bursting for load balancing between clouds).

[0049] A cloud computing environment is service oriented with a focus on statelessness, low coupling, modularity, and semantic interoperability. At the heart of cloud computing is an infrastructure that includes a network of interconnected nodes.

[0050]

[0023] Referring now to the figures, and in particular to Figures 1 and 2, which are exemplary diagrams of data processing environments in which example embodiments may be implemented. Figures 1 and 2 are merely exemplary and are not intended to assert or imply any limitation with respect to the environments in which different embodiments may be implemented. Particular embodiments may make many modifications to the depicted environments based on the following description.

[0051] 1 illustrates a block diagram of a network of data processing systems in which exemplary embodiments may be implemented. Data processing environment 100 is a network of computers in which exemplary embodiments may be implemented. Data processing environment 100 includes network 102. Network 102 is the medium used to provide communications links between various devices and computers connected together within data processing environment 100. Network 102 may include connections, such as wires, wireless communication links, or fiber optic cables.

[0052] Client or server are merely exemplary roles for certain data processing systems connected to network 102 and are not intended to exclude other configurations or roles of these data processing systems. Server 104 and server 106 connect to network 102 along with storage unit 108. Software applications may run on any computer in data processing environment 100. Clients 110, 112, and 114 are also connected to network 102. Data processing systems, such as servers 104 or 106, or clients 110, 112, or 114, may contain data and have software applications or software tools running thereon.

[0053] By way of example only, and without implying any limitation to such architecture, Figure 1 illustrates certain components that may be used in an exemplary implementation of an embodiment. For example, servers 104 and 106 and clients 110, 112, and 114 are illustrated as servers and clients by way of example only, and without implying any limitation to a client-server architecture. As another example, an embodiment may be distributed across multiple data processing systems and data networks as illustrated, while another embodiment may be implemented on a single data processing system within the scope of an exemplary embodiment. Data processing systems 104, 106, 110, 112, and 114 also represent exemplary nodes in clusters, partitions, and other configurations suitable for implementing an embodiment.

[0054] Device 132 is an example of a device as described herein. For example, device 132 may take the form of a smartphone, a tablet computer, a laptop computer, a stationary or portable client 110, a wearable computing device, or any other suitable device. Any software application described as executing in another data processing system in FIG. 1 may be configured to execute in device 132 in a similar manner. Any data or information stored or generated in another data processing system in FIG. 1 may be configured to be stored or generated in device 132 in a similar manner.

[0055] Application 105 implements the embodiments described herein. Application 105 executes on any of servers 104 and 106, clients 110, 112, and 114, and device 132. In addition, the analytics pipeline can execute on any of servers 104 and 106, clients 110, 112, and 114, and device 132, and does not have to execute on the same device as application 105.

[0056] The servers 104 and 106, the storage unit 108, and the clients 110, 112 and 114, and the device 132 may connect to the network 102 using wired connections, wireless communication protocols, or other suitable data connections. The clients 110, 112 and 114 may be, for example, personal computers or network computers.

[0057] In the illustrated example, server 104 may provide data, such as boot files, operating system images, and applications, to clients 110, 112, and 114. Clients 110, 112, and 114 may be clients to server 104 in this example. Clients 110, 112, 114, or some combination thereof, may include their own data, boot files, operating system images, and applications. Data processing environment 100 may include additional servers, clients, and other devices not shown.

[0058] In the depicted example, data processing environment 100 may be the Internet. Network 102 may represent a collection of networks and gateways that communicate with each other using Transmission Control Protocol / Internet Protocol (TCP / IP) and other protocols. At the heart of the Internet is a backbone of data communication links that route data and messages to major nodes or host computers, such as those encompassing thousands of commercial, government, educational, and other computer systems. Of course, data processing environment 100 may also be implemented as a number of different types of networks, such as an intranet, a local area network (LAN), or a wide area network (WAN). FIGURE 1 is intended as an example, and not as an architectural limitation for different illustrative implementations.

[0059] Among other uses, data processing environment 100 may be used to implement a client-server environment in which exemplary embodiments may be implemented. A client-server environment allows software applications and data to be distributed over a network such that applications function by using interactions between client and server data processing systems. Data processing environment 100 may also use a service-oriented architecture in which interoperable software components distributed across a network may be packaged together as coherent business applications. Data processing environment 100 may take the form of a cloud and may use a cloud computing model of service delivery to enable convenient, on-demand network access to a shared pool of configurable computing resources (e.g., networks, network bandwidth, servers, processing, memory, storage, applications, virtual machines, and services) that can be rapidly provisioned and released with minimal administrative effort or interaction with the provider of the service.

[0060] Referring to Figure 2, which illustrates a block diagram of a data processing system in which exemplary embodiments may be implemented. Data processing system 200 is an example of a computer, such as servers 104 and 106, or clients 110, 112, and 114 in Figure 1, or another type of device in which computer usable program code or instructions implementing the processes for the exemplary embodiments may be located.

[0061] Data processing system 200 is also representative of data processing systems or configurations thereof in which computer usable program code or instructions implementing the processes of the exemplary embodiments may be located, such as data processing system 132 in FIG. 1. Data processing system 200 is described as a computer by way of example only and not by way of limitation. Implementations in the form of other devices, such as device 132 in FIG. 1, may modify data processing system 200, for example, by adding a touch interface, or even omit certain illustrated components from data processing system 200, without departing from the general description of the operation and functionality of data processing system 200 described herein.

[0062] In the illustrated example, data processing system 200 uses a hub architecture, such as that described above, including a North Bridge and memory controller hub (NB / MCH) 202 and a South Bridge and input / output (I / O) controller hub (SB / ICH) 204. A processing unit 206, a main memory 208, and a graphics processor 210 are connected to North Bridge and memory controller hub (NB / MCH) 202. Processing unit 206 may include one or more processors and may be implemented using one or more heterogeneous processor systems. Processing unit 206 may be a multi-core processor. Graphics processor 210 may be connected to NB / MCH 202 via an accelerated graphics port (AGP) in some implementations.

[0063] In the illustrated example, a local area network (LAN) adapter 212 is connected to the south bridge and I / O controller hub (SB / ICH) 204. An audio adapter 216, a keyboard and mouse adapter 220, a modem 222, a read only memory (ROM) 224, a universal serial bus (USB) and other ports 232, and a PCI / PCIe device 234 are connected to the south bridge and I / O controller hub 204 through a bus 238. A hard disk drive (HDD) or solid state drive (SSD) 226 and a CD-ROM 230 are connected to the south bridge and I / O controller hub 204 through a bus 240. The PCI / PCIe device 234 may include, for example, an Ethernet adapter, an add-in card, and a PC card for a notebook computer. PCI uses a card bus controller, while PCIe does not. The ROM 224 may be, for example, a flash binary input / output system (BIOS). Hard disk drive 226 and CD-ROM 230 may use, for example, integrated drive electronics (IDE), serial advanced technology attachment (SATA) interfaces, or derivatives thereof, such as external SATA (eSATA) and micro SATA (mSATA). Super I / O (SIO) devices 236 may be connected to south bridge and I / O controller hub (SB / ICH) 204 via bus 238.

[0064] Memory, such as main memory 208, ROM 224, or flash memory (not shown), are some examples of computer usable storage devices. Hard disk drives or solid state drives 226, CD-ROM 230, and other similarly usable devices are some examples of computer usable storage devices, such as those mentioned above that encompass computer usable storage media.

[0065] An operating system runs on the processing unit 206. The operating system coordinates and provides control for various components within the data processing system 200 in FIG. 2. The operating system may be a commercially available operating system for any type of computing platform, including but not limited to server systems, personal computers, and mobile devices. An object-oriented or other type of programming system may work in conjunction with the operating system and may provide calls to the operating system from programs or applications running on the data processing system 200.

[0066] Instructions for the operating system, the object-oriented programming system, and applications or programs, such as application 105 in Figure 1, may be located on a storage device, such as in the form of code 226A on hard disk drive 226, and then loaded into at least one of one or more memories, such as main memory 208, for execution by processing unit 206. The processes of the exemplary implementations may be performed by processing unit 206 using computer-implemented instructions, which may be located in a memory, such as main memory 208, read-only memory 224, or in one or more peripheral devices.

[0067] Moreover, in one case, code 226A may be downloaded over network 201A from remote system 201B, where similar code 201C is stored on storage device 201D. In another case, code 226A may be downloaded over network 201A to remote system 201B, where downloaded code 201C is stored on storage device 201D.

[0068] The hardware in Figures 1-2 may vary depending on the implementation. Other internal hardware or peripheral devices, such as flash memory, equivalent non-volatile memory, or optical disk drives, may be used in addition to or in place of the hardware depicted in Figures 1-2. Additionally, the processes of the exemplary implementations may be applied to a multi-processor data processing system.

[0069] In some illustrative examples, data processing system 200 may be a personal digital assistant (PDA), which is typically configured with flash memory to provide non-volatile memory for storing operating system files or user generated data, or a combination thereof. The bus system may include one or more buses, such as a system bus, an I / O bus, a PCI bus, etc. Of course, the bus system may be implemented using any type of communications fabric or architecture that provides for a transfer of data between different components or devices connected to the fabric or architecture.

[0070] The communications unit may include one or more devices used to transmit and receive data, such as a modem or a network adapter. The memory may be, for example, main memory 208, or a cache, such as that found in north bridge and memory controller hub 202. The processing unit may include one or more processors or CPUs.

[0071] 1-2 and above-described examples are not meant to imply architectural limitations. For example, data processing system 200 may also be a tablet computer, a laptop computer, or a telephone device in addition to taking the form of a mobile or wearable device.

[0072] When a computer or data processing system is described as a virtual machine, virtual device, or virtual component, the virtual machine, virtual device, or virtual component operates in the manner of data processing system 200 using virtualized manifestations of some or all of the components depicted in data processing system 200. For example, in a virtual machine, virtual device, or virtual component, processing unit 206 is manifested as a virtualized instance of all or some of the hardware processing unit 206 available to the host data processing system, main memory 208 is manifested as a virtualized instance of all or some of the main memory 208 available to the host data processing system, and disk 226 is manifested as a virtualized instance of all or some of the disks 226 that may be available to the host data processing system. The host data processing system in such a case is represented by data processing system 200.

[0073] 3, which illustrates a block diagram of an exemplary configuration for automatically adjusting data access policies in data analytics, according to an exemplary embodiment. Application 300 is an example of application 105 in FIG. 1 and executes on any of servers 104 and 106, clients 110, 112 and 114, and device 132 in FIG.

[0074] Model parameter generation module 310 generates one or more autoencoder networks from one or more model parameters. In an implementation of module 310, the one or more model parameters are set to default values, set by a human expert based on knowledge of the data to be processed by the analytics operation, automatically generated, or set using another currently known technique, or set using a combination of these techniques.

[0075] A model training module 320 trains the generated autoencoder network according to one or more model parameters. Module 320 uses a set of original data and a set of data derived from the original data as training data. One implementation of module 320 uses the reconstruction error of the autoencoder network when processing a training set of original data as a loss function during training. Another implementation of module 320 uses the difference between an encoded form of the original training data (the output of the encoder part of the autoencoder network) and the corresponding derived training data as a loss function during training. Another implementation of module 320 trains the autoencoder network to minimize both loss functions.

[0076] Once module 320 has trained the generated autoencoder network for a specified number of training epochs, model evaluation module 330 measures the reconstruction error and complexity level of the autoencoder network. To measure the complexity level of the network, module 330 combines model parameters related to the complexity of the model into a common complexity level. Since both the number of hidden layers in each of the encoder and decoder parts and the number of dimensions of the encoded form of the data are measures of complexity, one implementation of module 330 normalizes the number of hidden layers in each of the encoder and decoder parts and the number of dimensions of the encoded form of the data to a common measure and measures the complexity level of the network by multiplying the two normalized values ​​together. Another implementation of module 330 also includes the number of training epochs used in training the specially configured autoencoder network in the complexity level measurement.

[0077] Module 330 uses the complexity level as a confidence score corresponding to the reconstruction error of the generated autoencoder network.

[0078] The access policy generation module 340 aggregates the reconstruction errors and corresponding complexity levels of one or more autoencoder networks into a level of difficulty score that represents a level of difficulty involved in recovering the initial data using the derived data output from the analytics operation.

[0079] In one implementation of application 300, module 330 determines a level of difficulty score for autoencoder networks generated from an initial set of model parameters, and module 310 varies one or more model parameters from their initial settings and generates a corresponding set of autoencoder networks. Module 320 trains the generated set of autoencoder networks according to each network's model parameters. Module 330 measures the reconstruction error and complexity level of each network, and module 340 aggregates the reconstruction error and corresponding complexity level of the autoencoder networks into a new level of difficulty score in a manner described herein.

[0080] The access policy generation module 340 uses the level of the difficulty score and the data access policy level of the initial data to determine the data access policy level of the derived data. For example, if the level of the difficulty score is low (e.g., 1 on a scale of 0 to 5), it indicates that it is not difficult at all to recover the initial data from the derived data, and the data access policy level of the derived data should be equal to the data access policy level of the initial data. On the other hand, if the level of the difficulty score is high (e.g., 4.5 on a scale of 0 to 5), it indicates that it is very difficult to recover the initial data from the derived data, and the data access policy level of the derived data can be lowered, i.e., relaxed, relative to the data access policy level of the initial data.

[0081] 4, which illustrates a block diagram of an exemplary configuration for automatically adjusting data access policies in data analytics, according to an exemplary embodiment. In particular, FIG. 4 illustrates the model evaluation module 330 in FIG. 3 in more detail.

[0082] The reconstruction error module 410 measures the reconstruction error of the autoencoder network. The complexity assessment module 420 measures the complexity level of the autoencoder network by combining model parameters related to the complexity of the model into a common complexity level. Since both the number of hidden layers in each of the encoder and decoder parts and the number of dimensions of the encoded form of the data are measures of complexity, one implementation of the module 420 normalizes the number of hidden layers in each of the encoder and decoder parts and the number of dimensions of the encoded form of the data to a common measure and measures the complexity level of the network by multiplying the two normalized values ​​together. Another implementation of the module 420 also includes the number of training epochs used in training the specially configured autoencoder network in the complexity level measurement.

[0083] 5, which illustrates an example of automatically adjusting data access policies in data analytics, according to an exemplary embodiment. This example can be implemented using application 300 in FIG.

[0084] As shown, input data 510 is subject to input data access policy 520. Input data access policy 520 includes a set of data access policy levels, with each data access policy level applied to a particular attribute or column of input data 510. For example, salary data for a particular individual may have a data access policy level of 4 (on a scale of 1 to 5) because the salary data requires a high level of protection, while the birth month of a particular individual may have a data access policy level of 1 (on a scale of 1 to 5) because, without additional data, the birth month requires a relatively low level of protection.

[0085] Input data 510 is processed by data transformation 530, resulting in derived data 540. Application 300 determines values ​​for policy levels in derived data access policies 550.

[0086] 6, which illustrates a continuous example of automatically adjusting data access policies in data analytics, according to an exemplary embodiment. The model parameter generation module 310 is the same as the model parameter generation module 310 in FIG. 3. The input data 510 is the same as the input data 510 in FIG. 5.

[0087] Input data 510 includes an input data column 602 that holds salary data. Application 300 determines derived data access policy values ​​for data derived from the data in input data column 602.

[0088] The model parameter generation module 310 generates a set of autoencoder networks by varying one or more model parameters. In particular, model 610 has no hidden layers. Model 620 has one hidden layer in each of the encoder and decoder portions and generates encoded data with a dimension of 50. Model 630 has one hidden layer in each of the encoder and decoder portions and generates encoded data with a dimension of 100. The depictions of models 610, 620, and 630 are symbolic illustrations of each model and are not intended to be exact representations of each model.

[0089] 7, which illustrates a continuous example of automatically adjusting data access policies in data analytics, according to an exemplary embodiment. Model training module 320 is the same as model training module 320 in FIG. 3. Models 610, 620, and 630 are the same as models 610, 620, and 630 in FIG. 6.

[0090] As shown, the model training module 320 trains a model 610 according to one or more model parameters to generate a trained model 710. The model training module 320 trains a model 620 according to one or more model parameters to generate a trained model 720. The model training module 320 trains a model 630 according to one or more model parameters to generate a trained model 730.

[0091] 8, which illustrates a continuing example of automatically adjusting data access policies in data analytics, according to an exemplary embodiment. The derived data 540 is the same as the derived data 540 in FIG. 5. The input data column 602 and the model 610 are the same as the input data column 602 and the model 610 in FIG. 6.

[0092] 8 illustrates in more detail the training of a model 610. One loss function used during training is a reconstruction loss 810, which is the reconstruction error of the model 610 when processing an input data column 602 to produce a reconstructed data column 802. Another loss function used during training is a loss function 840, which is the difference between a latent vector 830, which is the output of the encoder portion of the autoencoder network, and the derived features 820 in the derived data 540.

[0093] 9, which illustrates a continuous example of automatically adjusting data access policies in data analytics, according to an exemplary embodiment. The model evaluation module 330 is the same as the model evaluation module 330 in FIG. 3. The trained models 710, 720, and 730 are the same as the trained models 710, 720, and 730 in FIG. 7.

[0094] The model evaluation module 330 measures the reconstruction error and complexity level of each of the trained models 710, 720, and 730. In particular, the module 330 measures the reconstruction error 910 and complexity 912 of the trained model 710. The module 330 measures the reconstruction error 920 and complexity 922 of the trained model 720. The module 330 measures the reconstruction error 930 and complexity 932 of the trained model 730.

[0095] 10, the figure illustrates a continuous example of automatically adjusting data access policy in data analytics according to an exemplary embodiment. The access policy generation module 340 is the same as the access policy generation module 340 in FIG. 3. The reconstruction errors 910, 920, and 930 and the complexities 912, 922, and 932 are the same as the reconstruction errors 910, 920, and 930 and the complexities 912, 922, and 932 in FIG. 9.

[0096] The access policy generation module 340 aggregates the reconstruction errors 910, 920, and 930 and the complexities 912, 922, and 932 into an aggregated reconstruction error 1010 and confidence score 1012, determines a difficulty score level 1020, and uses the score 1020 and the data access policy level of the initial data to determine an access policy level 1030 for the derived data 540.

[0097] 11, which illustrates a flow chart diagram of an example process for automatically adjusting data access policies in data analytics, according to an example embodiment. The process 1100 can be implemented in the application 300 in FIG.

[0098] In block 1102, the application generates an autoencoder network from the model parameters. In block 1104, the application measures a reconstruction error and a complexity level for the autoencoder network. In block 1106, the application aggregates the reconstruction error and a confidence score corresponding to the complexity level into a level of difficulty score for the autoencoder network. In block 1108, the application generates a set of model parameters and corresponding autoencoder networks from the level of difficulty scores, each including a variation from the model parameters describing the autoencoder network. In block 1110, the application measures a model-specific reconstruction error and a corresponding complexity level for each of the set of autoencoder networks. In block 1112, the application aggregates the reconstruction error and a confidence score corresponding to the complexity level into a level of difficulty score for the set of autoencoder networks. In block 1114, the application generates a derived data access level from the level of difficulty score and an initial data access level corresponding to the initial set of data that provides access to a transformed data set generated by applying a transformation to the initial set of data. The application then terminates.

[0099] 12, an exemplary cloud computing environment 50 is illustrated. As illustrated, the cloud computing environment 50 includes one or more cloud computing nodes 10 with which local computing devices used by cloud consumers, such as a personal digital assistant (PDA) or mobile phone 54A, a desktop computer 54B, a laptop computer 54C, or an automobile computer system 54N, or combinations thereof, may communicate. The cloud computing nodes 10 may communicate with each other. They may be physically or virtually grouped in one or more networks, such as a private cloud, a community cloud, a public cloud, or a hybrid cloud, or combinations thereof, as described herein (not shown). This allows the cloud computing environment 50 to provide infrastructure, platform, or software, or combinations thereof, as a service without the cloud consumer having to maintain resources on the local computing device. It will be understood that the types of computing devices 54A-54N shown are intended to be illustrative only, and that the cloud computing node 10 and cloud computing environment 50 communicate with any type of computerized device (e.g., using a web browser) via any type of network or network-addressable connection or combination thereof.

[0100] 13, there is illustrated a set of functional abstraction layers provided by cloud computing environment 50 (FIG. 12). It should be understood that the components, layers, and functions are intended to be merely illustrative, and that embodiments of the present invention are not limited thereto. As illustrated, the following layers and corresponding functions are provided:

[0101] The hardware and software layer 60 includes hardware and software components. Examples of hardware components include mainframes 61, RISC (Reduced Instruction Set Computer) architecture based servers 62, servers 63, blade servers 64, storage devices 65, and networks and networking components 66. In some embodiments, the software components include network application server software 67 and database software 68.

[0102] The virtualization layer 70 provides an abstraction layer from which the following examples of virtual entities may be provided: virtual servers 71; virtual storage 72; virtual networks 73, including, for example, virtual private networks; virtual applications and operating systems 74; and virtual clients 75.

[0103] In one example, the management layer 80 may provide multiple functions described below. Resource provisioning 81 provides dynamic procurement of computing and other resources utilized to execute tasks within the cloud computing environment. Metering and pricing 82 provides cost tracking as resources are utilized within the cloud computing environment and billing or invoicing for the consumption of these resources. In one example, these resources may include application software licenses. Security provides identity verification for cloud consumers and tasks, and protection for data and other resources. User portal 83 provides access to the cloud computing environment for consumers and system administrators. Service level management 84 provides allocation and management of cloud computing resources such that required service levels are met. Service level agreement (SLA) planning and fulfillment 85 provides pre-provisioning and procurement of cloud computing resources where future requirements are predicted according to SLAs.

[0104] Workload tier 90 provides examples of functions for which the cloud computing environment may be utilized. Examples of workloads and functions that may be provided from this tier include mapping and navigation 91; software development and lifecycle management 92; virtual classroom instructional delivery 93; data analytics processing 94; transaction processing 95; and application selection based on a cumulative vulnerability risk assessment 96.

[0105] Thus, provided in example embodiments are computer-implemented methods, systems or apparatus, and computer program products for automatically adjusting data access policies in data analytics and other related features, functions or operations. Where an embodiment, or a portion thereof, is described in terms of a type of device, the computer-implemented method, system or apparatus, computer program product, or portions thereof, is adapted or configured for use with the appropriate and equivalent representation of that type of device.

[0106] When an embodiment is described as being implemented in an application, distribution of the application in a software as a service (SaaS) model is contemplated within the scope of the exemplary embodiment. In the SaaS model, the functionality of an application implementing an embodiment is provided to users by running the application in a cloud infrastructure. The users can access the application using various client devices through a thin client interface, such as a web browser (e.g., web-based email) or other lightweight client application. The users do not manage or control the underlying cloud infrastructure, including the network, servers, operating systems, or storage of the cloud infrastructure. In some cases, the users may not even manage or control the functionality of the SaaS application. In other cases, the SaaS implementation of the application may allow possible exceptions of limited user-specific application configuration settings.

[0107] The present invention may be a system, method, or computer program product, or combination thereof, at any level of technical detail that may be integrated. The computer program product may include one or more computer-readable storage media having computer-readable program instructions for causing a processor to perform aspects of the present invention.

[0108] The computer readable storage medium can be a tangible device capable of holding and storing instructions for use by an instruction execution device. The computer readable storage medium can be, for example, but not limited to, an electronic storage device, a magnetic storage device, an optical storage device, an electromagnetic storage device, a semiconductor storage device, or any suitable combination thereof. A non-exhaustive list of more specific examples of the computer readable storage medium includes the following: a portable computer diskette, a hard disk, a random access memory (RAM), a read only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), a static random access memory (SRAM), a portable compact disc read-only memory (CD-ROM), a digital versatile disk (DVD), a memory stick, a floppy disk, a mechanically encoded device such as a punch card or a ridge structure in a groove in which instructions are recorded, or any suitable combination thereof. As used herein, a computer-readable storage medium should not be construed as a transitory signal per se, such as an electric wave or other freely propagating electromagnetic wave, an electromagnetic wave propagating through a waveguide or other transmission medium (e.g., a light pulse passing through a fiber optic cable), or an electrical signal transmitted over an electrical wire.

[0109] The computer readable program instructions described herein can be downloaded from a computer readable storage medium to an individual computing device / processing device or to an external computer or external storage device via a network, such as the Internet, a local area network, a wide area network, or a wireless network, or a combination thereof. The network may be comprised of copper transmission cables, optical transmission fiber, wireless transmission, routers, firewalls, switches, gateway computers, or edge servers, or a combination thereof. A network adapter card or network interface in each computing device / processing device receives the computer readable program instructions from the network and transmits the computer readable program instructions to the individual computing device / processing device for storage in a computer readable storage medium.

[0110] The computer readable program instructions for carrying out the operations of the present invention may be either assembler instructions, instruction-set-architecture (ISA) instructions, machine instructions, machine-dependent instructions, microcode, firmware instructions, state setting data, configuration data for an integrated circuit, or source or object code written in any combination of one or more programming languages, such as object-oriented programming languages, such as Smalltalk, C++, and the like, conventional procedural programming languages ​​(e.g., the "C" programming language or similar programming languages). The computer readable program instructions may be executed entirely on the user's computer, partially on the user's computer, partially on the user's computer as a stand-alone software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or server. In the latter scenario, the remote computer may be connected to the user's computer via any kind of network, such as a local area network (LAN) or a wide area network (WAN), or the connection may be to an external computer (e.g., through the Internet using an Internet Service Provider). In some embodiments, electronic circuitry, such as programmable logic circuits, field-programmable gate arrays (FPGAs), or programmable logic arrays (PLAs), may execute computer-readable program instructions by utilizing state information of the computer-readable program instructions to personalize the electronic circuitry to perform aspects of the invention.

[0111] Aspects of the present invention are described herein with reference to flowchart illustrations or block diagrams, or combinations thereof, of methods, apparatus (systems), and computer program products or computer programs according to embodiments of the present invention. It will be understood that each block of the flowchart illustrations or block diagrams, or combinations thereof, and combinations of blocks in the flowchart illustrations or block diagrams, or combinations thereof, can be implemented by computer readable program instructions.

[0112] These computer readable program instructions may be provided to a processor of a computer or other programmable data processing apparatus, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, create means for implementing the functions / operations specified in one or more blocks of the flowchart illustrations or block diagrams, or combinations thereof, to produce a machine. These computer readable program instructions may also be stored in a computer readable storage medium that can instruct a computer, programmable data processing apparatus, or other device, or combinations thereof, to function in a particular manner, such that a computer readable storage medium having stored thereon includes an article of manufacture including instructions that implement aspects of the functions / operations specified in one or more blocks of the flowchart illustrations or block diagrams, or combinations thereof.

[0113] The computer readable program instructions may also be loaded onto a computer, other programmable data processing apparatus, or other device such that the instructions, which execute on a computer, other programmable data processing apparatus, or other device, implement the functions / acts identified in one or more blocks of the flowchart illustrations or block diagrams, or combinations thereof, to cause the computer, other programmable apparatus, or other device to perform a series of operational steps to generate a computer-implemented process.

[0114] The flowchart diagrams and block diagrams in the drawings illustrate the architecture, functionality, and operation of possible implementations of systems, methods, and computer program products or computer programs according to various embodiments of the present invention. In this regard, each block in the flowchart diagrams or block diagrams may represent a module, segment, or portion of an instruction, which includes one or more executable instructions for implementing one or more specified logical functions. In some alternative implementations, the functions shown in the blocks may occur out of the order shown in the drawings. For example, two blocks shown in succession may actually be accomplished as one step performed simultaneously, substantially simultaneously, partially or fully in a time-overlapping manner, depending on the functions involved, or the blocks may be performed in the reverse order. It should be noted that each block of the block diagrams or flowchart diagrams or combinations thereof, as well as combinations of multiple blocks of the block diagrams or flowchart diagrams or combinations thereof, may be implemented by a special purpose hardware-based system that performs the specified functions or operations, or may execute a combination of special purpose hardware and computer instructions.

Claims

1. 1. A computer-implemented method comprising: generating an autoencoder network from the first model parameters; measuring a reconstruction error for the autoencoder network, where the reconstruction error comprises a difference between an input to the autoencoder network and a corresponding output from the autoencoder network, the input to the autoencoder network comprising a portion of an initial set of data; aggregating the reconstruction error and a confidence score corresponding to a complexity level of the autoencoder network into a level of difficulty score of the autoencoder network; and generating a derived data access policy level corresponding to the initial data access policy level from the level of difficulty score and an initial data access policy level corresponding to the initial set of data, wherein the derived data access policy level provides access to a transformed data set generated by applying a transformation to the initial set of data. The method comprising:

2. training the autoencoder network using a training subset of the initial set of data. The computer-implemented method of claim 1 , further comprising:

3. The computer-implemented method of claim 2 , wherein the training is performed to minimize a reconstruction error of the autoencoder network.

4. 3. The computer-implemented method of claim 2, wherein the training is performed to minimize a difference between an output of an encoder portion of the autoencoder network and a transformed data set generated by applying the transformation to the training subset.

5. Measuring the complexity level for the autoencoder network. The computer-implemented method of claim 1 , further comprising:

6. generating a set of model parameters from the level of the difficulty score, where a second model parameter in the set of model parameters comprises a variation from the first model parameter; generating a set of autoencoder networks from the set of model parameters; measuring a model-specific reconstruction error of each autoencoder network in the set of autoencoder networks, where the model-specific reconstruction error comprises a difference between an input to an autoencoder network in the set of autoencoder networks and a corresponding output from the autoencoder network in the set of autoencoder networks, the input to the autoencoder network in the set of autoencoder networks comprising the portion of the initial set of data; and aggregating the model-specific reconstruction error and a set of confidence scores of each autoencoder network into a level of difficulty score in the set of autoencoder networks, where each confidence score corresponds to a complexity level of an autoencoder network in the set of autoencoder networks; The computer-implemented method of claim 1 , further comprising:

7. 2. The computer-implemented method of claim 1, wherein the model parameters include a number of hidden layers in an encoder portion of the autoencoder network and a number of hidden layers in a decoder portion of the autoencoder network.

8. The computer-implemented method of claim 1 , wherein the model parameters include a number of dimensions at the output of an encoder portion of the autoencoder network.

9. 1. A computer program for automatically adjusting a data access policy, the computer program comprising a plurality of program instructions collectively stored on one or more computer readable storage media, the plurality of program instructions comprising: program instructions for generating an autoencoder network from a first model parameter; program instructions for measuring a reconstruction error for the autoencoder network, where the reconstruction error comprises a difference between an input to the autoencoder network and a corresponding output from the autoencoder network, the input to the autoencoder network comprising a portion of an initial set of data; program instructions for aggregating the reconstruction error and a confidence score corresponding to a complexity level of the autoencoder network into a level of difficulty score of the autoencoder network; and and program instructions for generating, from the level of difficulty score and an initial data access policy level corresponding to the initial set of data, a derived data access policy level corresponding to the initial data access policy level, wherein the derived data access policy level provides access to a transformed data set generated by applying a transformation to the initial set of data. wherein the plurality of program instructions are executed by one or more processors. The computer program.

10. The plurality of program instructions: program instructions for training the autoencoder network using a training subset of the initial set of data. The computer program product of claim 9 , further comprising:

11. The computer program product of claim 10 , wherein the training is performed to minimize a reconstruction error of the autoencoder network.

12. 11. The computer program product of claim 10, wherein the training is performed to minimize a difference between an output of an encoder portion of the autoencoder network and a transformed data set generated by applying the transformation to the training subset.

13. The plurality of program instructions: program instructions for measuring the complexity level for the autoencoder network; The computer program product of claim 9 , further comprising:

14. The plurality of program instructions: program instructions for generating a set of model parameters from the level of the difficulty score, where a second model parameter in the set of model parameters comprises a variation from the first model parameter; program instructions for generating a set of autoencoder networks from the set of model parameters; program instructions for measuring a model-specific reconstruction error of each autoencoder network in the set of autoencoder networks, where the model-specific reconstruction error comprises a difference between an input to an autoencoder network in the set of autoencoder networks and a corresponding output from the autoencoder network in the set of autoencoder networks, the input to the autoencoder network in the set of autoencoder networks comprising the portion of the initial set of data; and and program instructions for aggregating the model-specific reconstruction error and a set of confidence scores of each autoencoder network into a level of difficulty score in the set of autoencoder networks, where each confidence score corresponds to a complexity level of an autoencoder network in the set of autoencoder networks. The computer program product of claim 9 , further comprising:

15. 10. The computer program product of claim 9, wherein the model parameters include a number of hidden layers in an encoder portion of the autoencoder network and a number of hidden layers in a decoder portion of the autoencoder network.

16. The computer program product of claim 9 , wherein the model parameters include a number of dimensions at an output of an encoder portion of the autoencoder network.

17. 10. The computer program product of claim 9, wherein the stored program instructions are stored within at least one of one or more storage media of a local data processing system, and the stored program instructions are transferred over a network from a remote data processing system.

18. 10. The computer program product of claim 9, wherein the stored program instructions are stored within at least one of one or more storage media of a server data processing system, and the stored program instructions are downloaded over a network to a remote data processing system for use in a computer readable storage device associated with the remote data processing system.

19. The computer program of claim 9 , wherein the computer program is provided as a service in a cloud environment.

20. 1. A computer system comprising: one or more processors; one or more computer readable memories; and one or more computer readable storage devices; and a plurality of program instructions stored on at least one computer readable storage device of the one or more computer readable storage devices for execution by at least one of the one or more processors via at least one of the one or more computer readable memories, the stored plurality of program instructions comprising: program instructions for generating an autoencoder network from a first model parameter; program instructions for measuring a reconstruction error for the autoencoder network, where the reconstruction error comprises a difference between an input to the autoencoder network and a corresponding output from the autoencoder network, the input to the autoencoder network comprising a portion of an initial set of data; program instructions for aggregating the reconstruction error and a confidence score corresponding to a complexity level of the autoencoder network into a level of difficulty score of the autoencoder network; and and program instructions for generating, from the level of difficulty score and an initial data access policy level corresponding to the initial set of data, a derived data access policy level corresponding to the initial data access policy level, wherein the derived data access policy level provides access to a transformed data set generated by applying a transformation to the initial set of data. The computer system comprising:

Images