A5063819347- Advanced Memory and Neural Computing
- Ferroelectric and Negative Capacitance Devices
- Neural Networks and Reservoir Computing
- Neural dynamics and brain function
- Semiconductor materials and devices
- Advancements in Semiconductor Devices and Circuit Design
- Quantum and electron transport phenomena
- Neural Networks and Applications
- Magnetic properties of thin films
- Machine Learning and ELM
- Neuroscience and Neural Engineering
- CCD and CMOS Imaging Sensors
- Photoreceptor and optogenetics research
- Graphene research and applications
- stochastic dynamics and bifurcation
- Surface and Thin Film Phenomena
- Domain Adaptation and Few-Shot Learning
- Semiconductor Quantum Structures and Devices
- Low-power high-performance VLSI design
- Carbon Nanotubes in Composites
- Quantum-Dot Cellular Automata
- Molecular Junctions and Nanostructures
- EEG and Brain-Computer Interfaces
- Nonlinear Dynamics and Pattern Formation
- Advanced Neural Network Applications
Centre de Nanosciences et de Nanotechnologies
2016-2025
Université Paris-Saclay
2016-2025
Centre National de la Recherche Scientifique
2016-2025
Université Paris-Sud
2013-2022
National Institute of Advanced Industrial Science and Technology
2019
Commissariat à l'Énergie Atomique et aux Énergies Alternatives
2011-2018
Micron (United States)
2018
National Yang Ming Chiao Tung University
2018
STMicroelectronics (Italy)
2018
CEA LETI
2018
Bioinspired hardware holds the promise of low-energy, intelligent, and highly adaptable computing systems. Applications span from automatic classification for big data management, through unmanned vehicle control, to control biomedical prosthesis. However, one major challenges fabricating bioinspired is building ultra-high-density networks out complex processing units interlinked by tunable connections. Nanometer-scale devices exploiting spin electronics (or spintronics) can be a key...
Memristive nanodevices can feature a compact multilevel nonvolatile memory function, but are prone to device variability. We propose novel neural network-based computing paradigm, which exploits their specific physics, and has virtual immunity devices used as synapses in spiking network performing unsupervised learning. They learn using simplified customized "spike timing dependent plasticity" rule. In the network, neurons' threshold is adjusted following homeostasis-type perform system...
Spin-transfer torque magnetic memory (STT-MRAM) is currently under intense academic and industrial development, since it features non-volatility, high write read speed endurance. In this work, we show that when used in a non-conventional regime, can additionally act as stochastic memristive device, appropriate to implement "synaptic" function. We introduce basic concepts relating spin-transfer tunnel junction (STT-MTJ, the STT-MRAM cell) behavior its possible use learning-capable synapses....
We demonstrate a unique energy efficient methodology to use Phase Change Memory (PCM) as synapse in ultra-dense large scale neuromorphic systems. PCM devices with different chalcogenide materials were characterized synaptic behavior. Multi-physical simulations used interpret the results. propose special circuit architecture ("the 2-PCM synapse"), read, write, and reset programming schemes suitable for of neural networks. A versatile behavioral model which can be simulating systems is...
In this paper, we present an alternative approach to neuromorphic systems based on multilevel resistive memory synapses and deterministic learning rules. We demonstrate original methodology use conductive-bridge RAM (CBRAM) devices as, easy program low-power, binary with stochastic New circuit architecture, programming strategy, probabilistic spike-timing dependent plasticity (STDP) rule for two different CBRAM configurations with-selector (1T-1R) without-selector (1R) are proposed. show...
Stochastic computing (SC), a radical rethinking of computation, defines operations on streams random bits; it trades precision for large advantages in speed. Implementation has been thwarted, though, by the lack an efficient means to properly decorrelate bitstreams at each logic gate SC circuit. This study harnesses recent advances manipulating magnetic skyrmions propose technique telegraph-signal reshuffling that is tailor-made applications. Leveraging two-dimensional diffusive character...
In neuroscience, population coding theory demonstrates that neural assemblies can achieve fault-tolerant information processing. Mapped to nanoelectronics, this strategy could allow for reliable computing with scaled-down, noisy, imperfect devices. Doing so requires the components form a set of basis functions in terms their response inputs, offering physical substrate computing. Such be implemented CMOS technology, but corresponding circuits have high area or energy requirements. Here, we...
Recent progress in artificial intelligence is largely attributed to the rapid development of machine learning, especially algorithm and neural network models. However, it performance hardware, particular energy efficiency a computing system that sets fundamental limit capability learning. Data-centric requires revolution hardware systems, since traditional digital computers based on transistors von Neumann architecture were not purposely designed for neuromorphic computing. A platform...
We introduce a novel energy-efficient methodology “2-PCM Synapse” to use phase-change memory (PCM) as synapses in large-scale neuromorphic systems. Our spiking neural network architecture exploits the gradual crystallization behavior of PCM devices for emulating both synaptic potentiation and depression. Unlike earlier attempts implement biological-like spike-timing-dependent plasticity learning rule with PCM, we simplified where long-term depression can be produced single invariant...
We propose a design methodology to exploit adaptive nanodevices (memristors), virtually immune their variability. Memristors are used as synapses in spiking neural network performing unsupervised learning. The memristors learn through an adaptation of spike timing dependent plasticity. Neurons' threshold is adjusted following homeostasis-type rule. System level simulations on textbook case show that performance can compare with traditional supervised networks similar complexity. They also...
Random number generation is critical for many emerging computing schemes, but the associated energy consumption and circuit area are major bottlenecks. This study exploits stochastic behavior of superparamagnetic tunnel junctions, magnetic nanodevices that, due solely to thermal noise, will switch randomly between two well-defined states. These junctions can produce high-quality, truly random bit streams, with an efficiency that orders magnitude better than state art. The authors furthermore...
In this work, we demonstrate an original methodology to use Conductive-Bridge RAM (CBRAM) devices as binary synapses in low-power stochastic neuromorphic systems. A new circuit architecture, programming strategy and probabilistic STDP learning rule are proposed. We show, for the first time, how intrinsic CBRAM device switching probability at ultra-low power can be exploited implement rule. Two complex applications demonstrated: real-time auditory (from 64-channel human cochlea) visual...
Cognitive tasks are essential for the modern applications of electronics, and rely on capability to perform inference. The Von Neumann bottleneck is an important issue such tasks, emerging memory devices offer opportunity overcome this by fusing computing memory, in nonvolatile instant on/off systems. A vision accomplishing use brain-inspired architectures, which excel at inference do not differentiate between memory. In work, we a neuroscience-inspired model learning, spike-timing-dependent...
In this work, we demonstrate how phase change memory (PCM) devices can be used to emulate biologically inspired synaptic functions in particular, potentiation and depression, important for implementing neuromorphic hardware. PCM with different chalcogenide materials are fabricated characterized. The asymmetry between the depression behaviors of is stressed. Detailed multi-physical simulations performed study underlying physics behavior PCM. A versatile behavioral model a multi-level...
Emerging non-volatile memories (NVM) based on resistive switching mechanism (RS) such as STT-MRAM, OxRRAM and CBRAM etc., are under intense R&D investigation by both academics industries. They provide high write/read speed, low power good endurance (e.g., > 10 <sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">12</sup> ) beyond mainstream NVMs, which allow them to be embedded directly with logic units for computing purpose. This integration could...
Spin-torque nano-oscillators can emulate neurons at the nanoscale. Recent works show that non-linearity of their oscillation amplitude be leveraged to achieve waveform classification for an input signal encoded in voltage. Here, we frequency and phase oscillator also used recognize waveforms. For this purpose, phase-lock waveform, which carries information its modulated frequency. In way, considerably decrease amplitude, phase, noise. We method allows classifying sine square waveforms with...
Owing to their nonvolatility, outstanding endurance, high write and read speeds, CMOS process compatibility, spin-transfer torque magnetoresistive memories (MRAMs) are prime candidates for innovative memory applications. However, the switching delay of core components-the magnetic tunnel junctions (MTJs)-is a stochastic quantity. To account this in electronic design, only partial models (working extreme regimes) available. In paper, we propose an analytical model current-driven MTJ, with...
Many outstanding studies have reported promising results in seizure forecasting, one of the most challenging predictive data analysis problems. This is mainly because electroencephalogram (EEG) bio-signal intensity very small, $\mu \text{V}$ range, and there are significant sensing difficulties given physiological non-physiological artifacts. Today process accurate epileptic identification labeling done by neurologists. The current unpredictability activities together with lack reliable...