Book chapter
Tensorizing neural networks
Deep neural networks currently demonstrate state-of-the-art performance in several domains.At the same time, models of this class are very demanding in terms of computational resources. In particular, a large amount of memory is required by commonly used fully-connected layers, making it hard to use the models on low-end devices and stopping the further increase of the model size. In this paper we convert the dense weight matrices of the fully-connected layers to the Tensor Train format such that the number of parameters is reduced by a huge factor and at the same time the expressive power of the layer is preserved.In particular, for the Very Deep VGG networks we report the compression factor of the dense weight matrix of a fully-connected layer up to 200000 times leading to the compression factor of the whole network up to 7 times.
In book
This book constitutes revised selected papers from the First International Workshop on Machine Learning, Optimization, and Big Data, MOD 2015, held in Taormina, Sicily, Italy, in July 2015. The 32 papers presented in this volume were carefully reviewed and selected from 73 submissions. They deal with the algorithms, methods and theories relevant in data science, optimization and machine learning.
The secular outcome of our investigation is development of new monitoring service for glucose control related to diabetes. It is based on the main results of research: 1) New innovative wearable sensor that carry non-invasive measurement of glucose level. Sensor uses several independent technologies, simultaneously: radio-frequency with different levels of signal, ultrasonic, electromagnetic and thermal; 2) Special mobile application as the principal interface monitor for personal usage; 3) The unique proprietary algorithm, based on neural net, which calculates the weighted average and returns the user's glucose level. The algorithm précises the results of measurements, based on genetic neural net ideas; 4) Special designed Data Base Storage in our cloud software specialized for gathering information and giving the predictions for patient. All results together makes the essence of the research.
Intelligent Systems Conference (IntelliSys) 2018 is the fourth research conference in the series. This conference is a part of SAI conferences being held since 2013. The conference series has featured keynote talks, special sessions, poster presentation, tutorials, workshops, and contributed papers each year. The conference focus on areas of intelligent systems and artificial intelligence (AI) and how it applies to the real world. IntelliSys is one of the best respected Artificial Intelligence (AI) Conference.
A lot of files and data, in general, are transferred throughout the networks. But the data may be corrupted by intrusions or package loss so, the executable files may be marked as non-executable and violate the local network policy. Thus, it’s necessary to detect such files. In this paper, we present a novel method for detecting broken bytes of a file, so the corrupted files may be detected. Also, the positions of wrong bytes might be helpful in restoring the original file content. This work is devoted to study of modern neural network models applied to detect corrupted bytes of a file problem. Since recurrent neural networks (RNNs) seem to be well suited for such tasks, the main tasks of this work are to analyze the efficiency of popular state-of-the-art RNNs solving the problem mentioned above and to compare results of different models. We use data consisting of the most popular file types collected from the Internet and manually randomly added noise to that data to test our models. An experiment on this data demonstrates the advantages and disadvantages of the considered models.
This book constitutes the proceedings of the 7th International Conference on Analysis of Images, Social Networks and Texts, AIST 2018, held in Moscow, Russia, in July 2018.
The 29 full papers were carefully reviewed and selected from 107 submissions (of which 26 papers were rejected without being reviewed). The papers are organized in topical sections on natural language processing; analysis of images and video; general topics of data analysis; analysis of dynamic behavior through event data; optimization problems on graphs and network structures; and innovative systems.
The relevance of scientific research in the field of quantum informatics is grounded. Highlighted promising areas of research. For foreign and Russian publications and materials, an overview of the main scientific results characterizing the current state of quantum informatics has been made. It is concluded that knowledge and resources are most intensively invested in the development of a quantum computer, its architecture and elements, quantum algorithms in the field of cryptography and artificial quantum intelligence. Developments are also actively underway in the field of modeling complex natural and artificial phenomena and processes, the application of quantum computing in the cognitive and social sciences
Working memory (WM) is a primary cognitive function that corresponds to the ability to update, stably maintain, and manipulate short-term memory (STM) rapidly to perform ongoing cognitive tasks. A prevalent neural substrate of WM coding is persistent neural activity, the property of neurons to remain active after having been activated by a transient sensory stimulus. This persistent activity allows for online maintenance of memory as well as its active manipulation necessary for task performance. WM is tightly capacity limited. Therefore, selective gating of sensory and internally generated information is crucial for WM function. While the exact neural substrate of selective gating remains unclear, increasing evidence suggests that it might be controlled by modulating ongoing oscillatory brain activity. Here, we review experiments and models that linked selective gating, persistent activity, and brain oscillations, putting them in the more general mechanistic context of WM. We do so by defining several operations necessary for successful WM function and then discussing how such operations may be carried out by mechanisms suggested by computational models. We specifically show how oscillatory mechanisms may provide a rapid and flexible active gating mechanism for WM operations.
This paper introduces the approach to the temperature parameters forecasting to avoid overheating of the spacecraft equipment at the end of the data transmission session during which information from temperature detectors are unavailable. It is essential to prevent situations in which spacecraft’s details will experience excessive overheating because it leads to the failure of major components. To determine the unknown temperature values of the spacecraft components based on current temperature mode and spaceship orientation parameters’ values at the indicated times algorithms for historiсal data processing are proposed. Preprocessing techniques were applied to the raw data accumulated during the operation period. The software for temperature parameters forecasting based on current orientation and temperature of the spacecraft’s components is provided. The conducted experiments proved the ability to reveal anomaly thermal situations and it was showed that the result error was small enough to accurately predict the possibility of overheating to prevent the spacecraft equipment failure.
Abstract:
The secular outcome of our investigation is development of new monitoring service for glucose control related to diabetes. It is based on the main results of research: 1) New innovative wearable sensor that carry non-invasive measurement of glucose level. Sensor uses several independent technologies, simultaneously: radio-frequency with different levels of signal, ultrasonic, electromagnetic and thermal; 2) Special mobile application as the principal interface monitor for personal usage; 3) The unique proprietary algorithm, based on neural net, which calculates the weighted average and returns the user's glucose level. The algorithm précises the results of measurements, based on genetic neural net ideas; 4) Special designed Data Base Storage in our cloud software specialized for gathering information and giving the predictions for patient. All results together makes the essence of the research.