Персона:
Алюшин, Александр Васильевич

Организационные подразделения

Организационная единица

Институт лазерных и плазменных технологий

Стратегическая цель Института ЛаПлаз – стать ведущей научной школой и ядром развития инноваций по лазерным, плазменным, радиационным и ускорительным технологиям, с уникальными образовательными программами, востребованными на российском и мировом рынке образовательных услуг.

Фамилия

Алюшин

Имя

Александр Васильевич

Полная страница элемента

Результаты поиска

Теперь показываю 1 - 10 из 10

Только метаданные
High-speed pattern matching architecture on limited connectivity FPGA
(2019) Arkhangelsky, V. G.; Alyushin, A. V.; Alyushin, S. A.; Алюшин, Александр Васильевич
© 2017 IEEE. State-of-the-art Field Programmable Gate Arrays (FPGA) are characterized by high frequency of operation, big volume and high bandwidth of its internal memory, which makes them very promising for rapid pattern matching systems realization. Large-scale data structures with high-level of internal connectivity of its elements require parallel data processing architectures with correspondent level of internal connectivity of its blocks. In this paper, we discuss different fields of application of the suggested by the authors approach to map one-dimensional input pattern to the two-dimensional FPGA processing structure with limited connectivity of its elements. Experimental and theoretical research have shown that synthesized on FPGA multicluster architectures for patterns with finely granulated structure matching are characterized by more than 90% utilization of FPGA memory bandwidth, low time delay of 3-5 clock cycles. For input patterns with coarse structure, effectiveness of FPGA memory elements use can decrease to 80%, data processing time delay increases to 5-6 clock cycles.
Только метаданные
Scalable reconfigurable memristive synaptic structures as a basis for the plasticity mechanisms in developing and self-organizing networks of artificial pulsed neurons
(2019) Arkhangelsky, V.; Alyushin, S.; Alyushin, A.; Алюшин, Александр Васильевич
© 2019 IEEE Of particular importance is the property of plasticity for networks of large dimension and scale, since the large dimension of the network is a necessary condition for the processes of self - organization and for "intellectual" data processing. However, the hardware implementation of this property is rather complicated. Using functional micro and nano-elements with memristive properties is a justified way of achieving the required high degree of reconfigurable synaptic structures integration and scale. Analysis of the chemical and biological mechanisms of STP and LTPl confirms the hypothesis of their relative independence and parallel learning of biological neural systems at different time scales. In the paper, the expediency of spatial and temporal scaling of artificial synaptic structures based on memristive elements, functioning on different physical principles, is proved.
Только метаданные
Memristive element functional model for computer based analysis and hardware emulation of pulsed neurons adaptive networks
(2019) Alyushin, S.; Arkhangelsky, V.; Alyushin, A.; Алюшин, Александр Васильевич
© 2019 IEEE The development of efficient and invariant to the integrated technology models for rapid design and behavioral analysis of new neural-like information processing systems of large dimensions is of particular significance. A special role in this process is played by functional models of memristive elements with computationally efficient implementation in modern CAD systems based on a limited number of library functional and electronic components (resistors - R, capacitors - C, active components such as OPAs, MOS structures and the like) and allowing simple hardware emulation. This paper presents a schematic model and an extended mathematical description of the functional memristor element with the structure 1C1R1MOS, proposed by the authors earlier. Now we describe its main electrical characteristics and operating modes. We also present the experimental study results of the memristor functional model in its hardware emulation as part of an artificial neural network.
Только метаданные
Acoustic Monitoring of the Psycho-Emotional State of Operational Personnel in the Management of High-Risk Objects Based on Neuromorphic Self-Learning Systems
(2020) Arkhangelsky, V. G.; Alyushin, A. V.; Alyushin, S. A.; Алюшин, Александр Васильевич
© 2020 IEEE.High confidence in the definition of operational personnel psycho-emotional stat is achieved by using a modern element base (processing environment)-a unified neuromorphic platform in the form of a fractal memristive structure with space-parametric self-learning. Experimental study of the proposed platform revealed its ability to synthesize new structural solutions and "dimensions"of the operational personnel psycho-emotional stat information representation.
Только метаданные
Structural and schematic methods of the biological neuron networks properties projecting on the architecture of modern integrated neuron-like systems with analog-digital information processing
(2019) Arkhangelsky, V.; Alyushin, S.; Alyushin, A.; Алюшин, Александр Васильевич
© 2019 IEEE This paper substantiates the essential role of space-time dependence of the information representation form in the development of intelleсtual energy-efficient and noise-resistant data processing systems of large dimensions. The authors propose the artificial neuron networks design approach with local analog and global digital neural data processing (LAGDNDP). According to LAGDNDP approach, we interpret the role of different forms of neural networks plasticity as the basis of global and local network adaptation to external conditions of functioning, internal parameters variation, network elements failure and degradation. We show the primary role of structural and circuit design methods in the development of highly integrated neuron-like architectures of large dimensions with advanced scalability, local and global adaptation. LAGDNDP approach together with modern integrated analog and digital technologies supports structural self-organization while training on the input data flows, the formation of a hierarchical fractal structure, emulation of multi-level spatial-temporal dynamics, sub-threshold information processing.
Только метаданные
Acoustic Monitoring of the Psycho-Emotional and Physical State of Operational Personnel in the Management of High-Risk Objects Based on Self-Learning Neuromorphic Systems
(2021) Alyushin, A. V.; Arkhangelsky, V. G.; Алюшин, Александр Васильевич
Только метаданные
Fractal Neuromorphic Pattern Recognition Architecture for Experimental Physics
(2021) Arkhangelskv, V. G.; Alyushin, A. V.; Алюшин, Александр Васильевич
© 2021 IEEE.A fractal neuromorphic architecture with repeated functional and structural properties of its elements at all levels of the internal hierarchy is proposed. The structure is focused on the search for rare useful events, 'pyramid' processing of large volumes of experimental data in real time with minimal time delay.
Только метаданные
Research and Development of Autonomous Neuromorphic Speech Stress Detector Based on Spike Representation of Information
(2020) Arkhangelsky, V. G.; Alyushin, S. A.; Alyushin, A. V.; Алюшин, Александр Васильевич
© 2020 IEEE.The paper presents the results of a comparative designs analysis of known neuromorphic FEE (Front End Electronics) for the primary adaptive time-frequency conversion of sound information into multidimensional streams of electrical signals as well as secondary data processing to identify characteristic features and determine the emotional component in the original speech signal based on analog CMOS circuits, switched capacitor circuits, digital FPGAs. The perspective directions of SSD (Speech Stress Detector) development on the basis of combined technologies of analog-digital CMOS and memristive elements are proved. The developed SSD prototype based on memristive analog-digital spike neural network is characterized by active FEE dynamics close to Hopf bifurcation, self-learning and adaptation in secondary information processing.
Только метаданные
Bit-Vector Pattern Matching Systems on the Basis of Analog-Digital Field Reprogrammable Arrays
(2020) Arkhangelsky, V. G.; Alyushin, S. A.; Alyushin, A. V.; Алюшин, Александр Васильевич
© 2020 IEEE.Bit-vector pattern matching technology is based on the effective decomposition of the primary digital multidimensional variables into corresponding groups of secondary variables of smaller dimension, followed by parallel processing the last of the above. The article proposes the use of the modern integrated technologies advantages for the implementation of efficient in terms of power consumption and hardware costs analog arithmetic-logic units with limited conversion accuracy for the development of promising analog-digital broadband bit-vector pattern matching systems (BVPMS). To improve the stability and accuracy of BVPMS, the paper substantiates the application of the locality principle for analog data processing, especially when using memristive elements with advanced functionality (non-volatile local storage of information, in-memory data processing, self-learning). The analysis of modern integrated memristive technologies is given and the directions of analog-digital field reprogrammable arrays realization for BVMP are proved.).
Только метаданные
Memristive Element with Multiple Internal State Variables Functional Model for Computer Based Analysis and Hardware Emulation of Pulsed Neural Adaptive Networks
(2020) Alyushin, S. A.; Arkhangelsky, V. G.; Alyushin, A. V.; Алюшин, Александр Васильевич
© 2020 IEEE.A functional model of a memristive element (ME) with multiple internal state variables (ISV) is proposed. The dynamics of each of the ME ISV is determined by functional transformations of the ME input stimulus by the corresponding integral and differential form and time constant. ME functional model can be efficiently implemented in modern CAD systems based on a limited number of library components and allowing simple hardware emulation. Theoretical and experimental study of the proposed model with two ISVs and its hardware emulation for the mimicry of the sodium ion channel in the development of pulse neuron adaptive networks showed its adequate response to external influences, similar to the Hodgkin-Huxley model. Truncations of this model to the first order (1C1R1MOS) with one ISV exhibit the property of forward and reverse memristivity, can be applied to mimicry of potassium ion channels.