32 results about "Program Design Language" patented technology

The disclosed system provides a transformation-based implementation of forward-mode and reverse-mode automatic differentiation as a built-in, first-class function in a functional programming language. Each of these constructs imposes only a small constant factor of the computational burden (time) of the function itself, and the forward construct has the same properties in terms of space. The functions can be applied to any function, including those involving derivatives and nested closures.

The present invention relates to an electronic device, a compiling method, and a computer-readable recording medium. An electronic device according to an embodiment of the present invention compiles a source code of a dynamic programming language compiled during execution of a program to the level of a static programming language compiled before execution, and may comprise a processor which determines whether the source code of the dynamic programming language generated when a command for operating the program is received can be processed by the first compiler in the same manner as the source code of the static programming language processing, and converts the source code of the dynamic programming language by the first compiler if the code can be processed as a result of the determination.

An approach to computational processing and programming calls for the creation of multiple software entities ("molecules 10") that can be dynamically configured to autonomously process data. Molecules can send and receive signals that indicate the state of the molecule, whether the processing task is complete, the result of the processing task, and whether to terminate, interrupt, reconfigure, or continue the molecule through the creation of one or more "next" molecules. operate. Molecules can be created from a common set of software microcomponents (40, 42, 44, 46, 48), which can be written in any programming language to run in any operating system environment. Molecules can reside with a single computing resource, however, they are ideally suited for deployment with different types of distributed computing resources on a network or in a parallel processing environment. All data processing tasks are performed by creating molecules in "logical networks" that are dynamically adapted to accomplish their tasks. Logical networks can be allocated to work with different legacy systems, and applications programmed in different languages, and with any type of data stored in any format. As a result, data processing tasks in distributed or parallel processing environments can be performed more efficiently and entirely new types of network computing or parallel processing tasks can be adopted.

The present invention proposes a standardized training platform and automation system for artificial intelligence models. The training platform and the automation system are developed based on Python (programming language) and deployed on a platform with a GPU (graphics processing engine) using Docker (application container engine) virtualization technology. Device) server, using TCP / IP (Internet Transfer Protocol) protocol for communication, applied in the field of machine learning, supporting mainstream deep learning (Deep Learning, DL) TensorFlow, PyTorch, Caffe, Keras, Theano and other frameworks; users only need to Click the button on the client or use some script commands to automatically upload the code, download the data and use the GPU (Graphics Processing Unit) to train your own model, and complete the model evaluation and automatic optimization iterations; in the whole process of training the model, the present invention A set of operation system is provided to ensure the automation, standardization and specialization of the artificial intelligence model training process, saving time and simplifying steps.

The invention provides a unified data service platform and an implementation method of the unified data service platform. The implementation method of the unified data service platform comprises the steps that an operation request from an application server is received through an RESTful interface, wherein the operation request includes a data object and an operation method; a corresponding and executable operation process is searched for in a memory according to the data object and the operation method in the operation request, and a database server where the data object in the operation request is located is searched for according to the preset mapping relation between the data object and the database server; the searched executable operation process is sent to the database server to be executed. According to the unified data service platform and the implementation method of the unified data service platform, the purpose that a data service is irrelevant to a specific program design language can be achieved, high extensibility is achieved, and effectiveness of the data service is improved.

A transmission module for a machine control, is designed to process an input program code of a control program, and based thereon, to generate a uniform intermediate code having work instructions for execution by the machine control. The input program code comprises a first program code component and a second program code component, where the first program code component is written in a first programming language and the second program code component is written in a second programming language. The work instructions of the intermediate code comprise a first work instruction and a second work instruction, where the first work instruction is generated on the basis of the first program code component and the second work instruction is generated on the basis of the second program code component.

The present invention is a kind of multi-task compiling method based on graphical guard-style command calculus, relates to a graphical guard-style command calculus GCC_graph automatic compiling method to generate executable multi-task Ada / C code, for the GCC_graph program input by the designer, the The compiler automatically compiles it to generate the corresponding executable multi-task Ada / C code. All the functions of the compiler are implemented by the functional programming language CAML programming. Since CAML is a safe language, the CAML program is verified by its own compiler before execution. In addition, the compiler designed by the present invention uses a modular structure, once the program changes, we only need to modify the code of the corresponding module.

The present invention is applicable to the field of computer technology, and provides a block chain-based file protection method and terminal equipment. The method includes: obtaining type information of a first file to be protected; the type information is used to identify the first The programming language used by the source file corresponding to the file; determine the protection information of the first file according to the type information; the protection information is used to prevent decompilation; configure the protection information to the The first file obtains a second file; the execution logic of the second file is the same as the execution logic of the first file. The present invention can improve the data security of the first file and improve the corresponding The purpose of the security of the application's user information or user property.

The invention provides a method for automatically generating function descriptive annotations for program source codes, comprising: step 1: extracting features of source code; step 2: analyzing the extracted features, and automatically expressing features about functions in natural language. The invention can automatically generate corresponding function descriptive comments for various programming languages, well expresses the functions and functions of source codes, and can effectively promote software maintenance and development. Compared with the prior art, the present invention has the following beneficial effects: 1. Compared with other methods, the present invention can characterize the characteristics of the source code more comprehensively by processing the parse tree of the source code and extracting features, not only 2. The present invention no longer relies on fixed templates when generating annotations. Compared with other methods, it is more flexible and diverse, and the generated annotations can also express more functions, and the description is more comprehensive; 3. The present invention is applicable to Various programming languages ​​are universal.