198 results about "Learning rule" patented technology

Content segmentation, categorization and identification methods are described. Content tracking approaches are illustrated that are suitable for large scale deployment. Time-aligned applications such as multi-language selection, customized advertisements, second screen services and content monitoring applications can be economically deployed at large scales. A client performs fingerprinting, scene change detection, audio turn detection, and logo detection on incoming video and gathers database search results, logos and text to identify and segment video streams into content, promos, and commercials. A learning engine is configured to learn rules for optimal identification and segmentation at each client for each channel and program. Content sensed at the client site is tracked with reduced computation and applications are executed with timing precision. A user interface for time-aligned publishing of content and subsequent usage and interaction on one or more displays is also described.

Generalized learning rules may be implemented. A framework may be used to enable adaptive spiking neuron signal processing system to flexibly combine different learning rules (supervised, unsupervised, reinforcement learning) with different methods (online or batch learning). The generalized learning framework may employ time-averaged performance function as the learning measure thereby enabling modular architecture where learning tasks are separated from control tasks, so that changes in one of the modules do not necessitate changes within the other. Separation of learning tasks from the control tasks implementations may allow dynamic reconfiguration of the learning block in response to a task change or learning method change in real time. The generalized spiking neuron learning apparatus may be capable of implementing several learning rules concurrently based on the desired control application and without requiring users to explicitly identify the required learning rule composition for that task.

Apparatus and methods for universal node design implementing a universal learning rule in a mixed signal spiking neural network. In one implementation, at one instance, the node apparatus, operable according to the parameterized universal learning model, receives a mixture of analog and spiking inputs, and generates a spiking output based on the model parameter for that node that is selected by the parameterized model for that specific mix of inputs. At another instance, the same node receives a different mix of inputs, that also may comprise only analog or only spiking inputs and generates an analog output based on a different value of the node parameter that is selected by the model for the second mix of inputs. In another implementation, the node apparatus may change its output from analog to spiking responsive to a training input for the same inputs.

Neural networks for optimal estimation (including prediction) and / or control involve an execution step and a learning step, and are characterized by the learning step being performed by neural computations. The set of learning rules cause the circuit's connection strengths to learn to approximate the optimal estimation and / or control function that minimizes estimation error and / or a measure of control cost. The classical Kalman filter and the classical Kalman optimal controller are important examples of such an optimal estimation and / or control function. The circuit uses only a stream of noisy measurements to infer relevant properties of the external dynamical system, learn the optimal estimation and / or control function, and apply its learning of this optimal function to input data streams in an online manner. In this way, the circuit simultaneously learns and generates estimates and / or control output signals that are optimal, given the network's current state of learning.

Generalized learning rules may be implemented. A framework may be used to enable adaptive signal processing system to flexibly, combine different learning rules (supervised, unsupervised, reinforcement learning) with different methods (online or batch learning). The generalized learning framework may employ time-averaged performance function as the learning measure thereby enabling modular architecture where learning tasks are separated from control tasks, so that changes in one of the modules do not necessitate changes within the other. The generalized learning apparatus may be capable of implementing several learning rules concurrently based on the desired control application and without requiring users to explicitly identify the required learning rule composition for that application.

A method and system is provided in which data packets from multiple customer VLANs are forwarded over a MAN using VLAN aggregation. A layer-2 switch located at the edge of the MAN connects the customer VLANs to the MAN. The edge switch aggregates multiple customer VLANs (the “sub-VLANs”) into one provider VLAN (the “super-VLAN”). When a packet is forwarded from the sub-VLAN to the super-VLAN and vice versa, the edge switch uses modified bridge forwarding rules to exchange the customer-configured VLAN-IDs with the provider-configured VLAN-IDs before transporting the packet over the MAN. The edge switch further uses modified bridge media access control (MAC) address learning rules to isolate one customer's traffic from another's (i.e. isolate one sub-VLAN's traffic from another sub-VLAN's traffic).

An information processing system having neuron-like signal processors that are interconnected by synapse-like processing junctions that simulates and extends capabilities of biological neural networks. The information processing systems uses integrate-and-fire neurons and Temporally Asymmetric Hebbian learning (spike timing-dependent learning) to adapt the synaptic strengths. The synaptic strengths of each neuron are guaranteed to become optimal during the course of learning either for estimating the parameters of a dynamic system (system identification) or for computing the first principal component. This neural network is well-suited for hardware implementations, since the learning rule for the synaptic strengths only requires computing either spike-time differences or correlations. Such hardware implementation may be used for predicting and recognizing audiovisual information or for improving cortical processing by a prosthetic device.

The present invention relates to a method of risk analysis in an automatic intrusion response system that provides computer-related security in a large scale dynamic network environment, comprising: (a) classifying intrusion detection information by using IDMEF data model; (b) establishing a risk assessment knowledge base; (c) learning rules of said knowledge base; and (d) assessing the risk level of an external attack based upon said knowledge base. Said risk level is determined by parameters such as intrusion detection information, weakness information, network bandwidth, system performance and importance, and frequency of attacks, etc.

A system and method for separating a mixture of audio signal into desired audio signals (430) (e.g., speech) and a noise sign (440) is disclosed. Microphones (310, 320) are positioned to receive the mixed audio signals, and an independent component analysis (ICA) processes (212) the sound mixture using stability constraints. The ICA process (508) uses predefined characteristics of the desired speech signal to identify and isolate a target sound signal (430). Filter coefficients are adapted with a learning rule and filter weight update dynamics are stabilized to assist convergence to a stable separated ICA signal result. The separated signals may be peripherally-processed to further reduce noise effects using post-processing (214) and pre-processing (220, 230) techniques and information. The proposed system is designed and easily adaptable for implementation on DSP units or CPUs in audio communication hardware environments.

The invention discloses an intelligent fault classification and location method for an ultra-high voltage direct current transmission line, and belongs to the technical field of relay protection of power systems. The method comprises the following steps of: classifying fault data by using a neural network by adopting a layered and distributed neural network model; distinguishing fault types; sending the classified data into different neural networks respectively for performing fault location; when the direct current transmission line has a fault and a sampling frequency is 10 kHz, selecting a discrete line mode voltage signal which has the sampling sequence length of 100 after the fault and performing S-transform, wherein a transform result is a complex time-frequency matrix of 51*100; solving the modulus of each element in the complex matrix to obtain transient energy distribution of the line mode voltage at all frequencies; selecting first five energy spectrums as sample properties; selecting a transfer function and a learning rule; setting proper neural network parameters for constructing a BP network model; and performing fault classification and fault location. A large number of simulation results show that the method has a good effect.

Apparatus and methods for universal node design implementing a universal learning rule in a mixed signal spiking neural network. In one implementation, at one instance, the node apparatus, operable according to the parameterized universal learning model, receives a mixture of analog and spiking inputs, and generates a spiking output based on the model parameter for that node that is selected by the parameterized model for that specific mix of inputs. At another instance, the same node receives a different mix of inputs, that also may comprise only analog or only spiking inputs and generates an analog output based on a different value of the node parameter that is selected by the model for the second mix of inputs. In another implementation, the node apparatus may change its output from analog to spiking responsive to a training input for the same inputs.

A method for protecting a Web application running on a first local Web Server bases from hacker attacks, said Web Server being connectable to at least one client, the method comprising the following steps: —providing a plurality of preset rules on said Server, which correspond to specific characteristics of HTTP requests; —receiving an HTTP request on said server from the client, said HTTP request comprising a plurality of characteristics; —analyzing said characteristics of said received HTTP request in accordance with said rules provided on said server; —rejecting said HTTP request, if said rules identify said HTTP request as harmful request; —accepting said HTTP request, if said rules identify said HTTP request as trustable request; —classifying said HTTP request as doubtful request, if said rules identify said request neither as harmful request nor as trustable request; —evaluating the characteristics of said doubtful local request; —generating a learned rule on basis of the edge base evaluation.

The invention discloses a building air-conditioning energy consumption prediction method based on a BP neural network model. The method comprises: analyzing influence factors of building air-conditioning energy consumption; according to influence parameters, collecting historical building air-conditioning energy consumption sample parameters, and preprocessing the parameters; using a BP neural network, according to dimensionality of the sample parameters, establishing a building air-conditioning energy consumption prediction model; using the preprocessed sample parameters as a training sample, training the building air-conditioning energy consumption prediction model; collecting near-term real-time building air-conditioning energy consumption sample parameters to evaluate the building air-conditioning energy consumption prediction model; if errors are in an allowed range, output of the model being a building air-conditioning energy consumption predicted value; if not, training the model again. The building air-conditioning energy consumption prediction method based on a BP neural network model is advantaged in that learning rules are simple, a computer can easily implement, and the method has excellent robustness, memory capability, nonlinear mapping capability, and powerful self-learning capability.

A dynamic bandwidth allocation method of an Ethernet passive optical network, comprises a predictor and a rule of QoS-promoted dynamic bandwidth allocation (PQ-DBA); the predictor predicts a client behavior and numbers of various kinds of packets by using a pipeline scheduling predictor consisted of a pipelined recurrent neural network (PRNN), and a learning rule of the extended recursive least squares (ERLS); the present invention establishes a better QoS traffic management for the OLT-allocated ONU bandwidth and client packets sent by priority.

Neural networks for optimal estimation (including prediction) and / or control involve an execution step and a learning step, and are characterized by the learning step being performed by neural computations. The set of learning rules cause the circuit's connection strengths to learn to approximate the optimal estimation and / or control function that minimizes estimation error and / or a measure of control cost. The classical Kalman filter and the classical Kalman optimal controller are important examples of such an optimal estimation and / or control function. The circuit uses only a stream of noisy measurements to infer relevant properties of the external dynamical system, learn the optimal estimation and / or control function, and apply its learning of this optimal function to input data streams in an online manner. In this way, the circuit simultaneously learns and generates estimates and / or control output signals that are optimal, given the network's current state of learning.

Apparatus and methods for universal node design implementing a universal learning rule in a mixed signal spiking neural network. In one implementation, at one instance, the node apparatus, operable according to the parameterized universal learning model, receives a mixture of analog and spiking inputs, and generates a spiking output based on the model parameter for that node that is selected by the parameterized model for that specific mix of inputs. At another instance, the same node receives a different mix of inputs, that also may comprise only analog or only spiking inputs and generates an analog output based on a different value of the node parameter that is selected by the model for the second mix of inputs. In another implementation, the node apparatus may change its output from analog to spiking responsive to a training input for the same inputs.

Generalized learning rules may be implemented. A framework may be used to enable adaptive signal processing system to flexibly combine different learning rules (supervised, unsupervised, reinforcement learning) with different methods (online or batch learning). The generalized learning framework may employ average performance function as the learning measure thereby enabling modular architecture where learning tasks are separated from control tasks, so that changes in one of the modules do not necessitate changes within the other. Separation of learning tasks from the control tasks implementations may allow dynamic reconfiguration of the learning block in response to a task change or learning method change in real time. The generalized learning apparatus may be capable of implementing several learning rules concurrently based on the desired control application and without requiring users to explicitly identify the required learning rule composition for that application.

The embodiment of the invention discloses an alarm processing method and device, computer equipment and a storage medium. The method comprises the steps of monitoring monitor data of a monitor object;according to preset alarm rules, determining alarm data in the monitor data, wherein the alarm data is the monitor data triggering the preset alarm rules; according to the alarm data, generating an alarm event, and recording the alarm event into a preset message queue; according to the preset obtained rules, obtaining the alarm event in the preset message queue; on the basis of machine learning rules and a preset knowledge base, preprocessing the alarm event; sending the preprocessed alarm event to a user terminal. According to the alarm processing method, alarm storm can be eliminated, meanwhile, the alarm accuracy can be improved, and the number of invalid alarms is decreased.

The invention provides a dynamic spectrum access method on the basis that the policy planning restricts Q learning, which comprises the following steps: cognitive users can divide the frequency spectrum state space, and select out the reasonable and legal state space; the state space can be ranked and modularized; each ranked module can finish the Q form initialization operation before finishing the Q learning; each module can individually execute the Q learning algorithm; the algorithm can be selected according to the learning rule and actions; the actions finally adopted by the cognitive users can be obtained by making the strategic decisions by comprehensively considering all the learning modules; whether the selected access frequency spectrum is in conflict with the authorized users is determined; if so, the collision probability is worked out; otherwise, the next step is executed; whether an environmental policy planning knowledge base is changed is determined; if so, the environmental policy planning knowledge base is updated, and the learning Q value is adjusted; the above part steps are repeatedly executed till the learning convergence. The method can improve the whole system performance, and overcome the learning blindness of the intelligent body, enhance the learning efficiency, and speed up the convergence speed.

The invention relates to a structured report template-based medical image knowledge base establishment method and belongs to the field of medical information processing and application. For improving data availability, a multi-modal stored medical image is converted into structured big data, so that data innovation and value addition are realized and accurate data support is provided for medical research. The method comprises the steps of establishing characteristic noumena of the medical image; defining standards of disease syndromes; designing report templates by taking structured data as elements; forming knowledge elements and exchanging information; and establishing learning rules. Through noumenon library establishment, disease syndrome definition, a text report template, an image-text report template and a CDA document, a relationship between a diagnosis conclusion and image presentation is established, so that the retrieval efficiency is high and the accuracy is greatly enhanced; by establishing the learning rules, the knowledge elements are continuously perfected, so that the medical research is assisted and the medical image level of the whole industry is improved; and the method has extremely wide application values.

The invention discloses a weight adjustment circuit for variable-resistance synapses, which relates to the fields of integrated circuits and neural networks, and is used for carrying out weight adjustment on variable-resistance synapses. The circuit is composed of a weight enhancement adjustment subcircuit A (LTP (long term potentiation) adjustment) and a weight inhibition adjustment subcircuit B (LTD (long term depression) adjustment), wherein the two subcircuits respectively contain a charging pole, a discharging pole, a charge storage pole and an output pole. The core of the circuit is implemented by using an analog circuit mode, therefore, the number of transistors required by the circuit is greatly reduced; and meanwhile, through the setting of the bias voltage on a discharge tube in the discharge pole, the size of a weight adjustment time window can be adjusted conveniently. The circuit disclosed by the invention follows an STDP (spike timing dependent plasticity) learning rule, and LTP and LTD pulse outputs are generated according to the activities of nerve units at the two ends of the variable-resistance synapses so as to carry out corresponding weight adjustment on the variable-resistance synapses. The circuit disclosed by the invention is simple in structure, convenient in parameter adjustment, and suitable for applications, such as weight adjustment on electronic synapses of a large-scale neural network, and the like.

A system adapted to implement a learning rule in a three-dimensional (3D) environment is described. The system includes: a renderer adapted to generate a two-dimensional (2D) image based at least partly on a 3D scene; a computational element adapted to generate a set of appearance features based at least partly on the 2D image; and an attribute classifier adapted to generate at least one set of learned features based at least partly on the set of appearance features and to generate a set of estimated scene features based at least partly on the set of learned features. A method labels each image from among the set of 2D images with scene information regarding the 3D scene; selects a set of learning modifiers based at least partly on the labeling of at least two images; and updates a set of weights based at least partly on the set of learning modifiers.

A system and method for separating a mixture of audio signal into desired audio signals (430) (e.g., speech) and a noise sign (440) is disclosed. Microphones (310, 320) are positioned to receive the mixed audio signals, and an independent component analysis (ICA) processes (212) the sound mixture using stability constraints. The ICA process (508) uses predefined characteristics of the desired speech signal to identify and isolate a target sound signal (430). Filter coefficients are adapted with a learning rule and filter weight update dynamics are stabilized to assist convergence to a stable separated ICA signal result. The separated signals may be peripherally-processed to further reduce noise effects using post-processing (214) and pre-processing (220, 230) techniques and information. The proposed system is designed and easily adaptable for implementation on DSP units or CPUs in audio communication hardware environments.

The invention discloses a coal mine accident simulating method and system based on a multi-intelligent agent. The method comprises the following steps: constructing an intelligent agent entity and a virtual simulating environment; selecting a preset model and preset knowledge to perform intelligent agent model computation to form a model computation result; performing the intelligent agent entity action computation and selecting the intelligent agent entity behavior; using the intelligent agent action experience learning rule to perform the intelligent agent entity behavior configuration, executing the intelligent agent entity behavior, updating the environment variable of the virtual simulating environment, and updating the current virtual simulating environment by a simulating scene constructing module based on the intelligent agent behavior state information and the updated environment variable; and updating a behavior selection parameter, and using the updated behavior selection parameter to perform the intelligent agent entity behavior computation at the next moment. The method and system provided by the invention can be used for imitating coal mine accidents in various environments so as to provide abundant information for immediately discovering the coal mine accident potential.

There are equipped a first recurrent neural network formed by connecting plural nodes so as to have a loop in which the output from one node is input to another node in accordance with a predetermined coupling weight coefficient. Meanwhile, the output of at least one node is fed back to the node concerned or another node, and an optimizing unit for determining the optimum solution of the coupling weight coefficient in the first recurrent neural network based on a learning rule using a hereditary algorithm. In this case, the first recurrent neural network outputs a first parameter indicating a motion state of a vehicle based on predetermined input information, thereby functioning as a vehicle motion model.

A method for protecting a Web application running on a first local Web Server bases from hacker attacks, said Web Server being connectable to at least one client, the method comprising the following steps: —providing a plurality of preset rules on said Server, which correspond to specific characteristics of HTTP requests; —receiving an HTTP request on said server from the client, said HTTP request comprising a plurality of characteristics; —analyzing said characteristics of said received HTTP request in accordance with said rules provided on said server; —rejecting said HTTP request, if said rules identify said HTTP request as harmful request; —accepting said HTTP request, if said rules identify said HTTP request as trustable request; —classifying said HTTP request as doubtful request, if said rules identify said request neither as harmful request nor as trustable request; —evaluating the characteristics of said doubtful local request; —generating a learned rule on basis of the edge base evaluation.

Apparatus and methods for universal node design implementing a universal learning rule in a mixed signal spiking neural network. In one implementation, at one instance, the node apparatus, operable according to the parameterized universal learning model, receives a mixture of analog and spiking inputs, and generates a spiking output based on the model parameter for that node that is selected by the parameterized model for that specific mix of inputs. At another instance, the same node receives a different mix of inputs, that also may comprise only analog or only spiking inputs and generates an analog output based on a different value of the node parameter that is selected by the model for the second mix of inputs. In another implementation, the node apparatus may change its output from analog to spiking responsive to a training input for the same inputs.

The invention discloses an unsupervised learning method of handwritten digit identification. The method comprises the following steps: firstly, establishing a spiking neural network model, and determining the connection mode between neurons according to an analog biological information processing mechanism; secondly, establishing a spiking time dependent plasticity model according to the synaptictrace learning rules; and finally, optimizing the pre-synaptic initial weight and the synaptic delay in the spiking neural networks by using a genetic algorithm, training and testing the MNIST (MixedNational Institute of Standards and Technology database) data set, thus achieving the unsupervised learning of handwritten digit identification. According to the unsupervised learning method of handwritten digit identification, experiments show that the training speed is effectively improved and the recognition accuracy is improved.