Jean-Michel Portal

Bio: Jean-Michel Portal is an academic researcher from Centre national de la recherche scientifique. The author has contributed to research in topics: Resistive random-access memory & Artificial neural network. The author has an hindex of 25, co-authored 136 publications receiving 2047 citations. Previous affiliations of Jean-Michel Portal include Alternatives & Aix-Marseille University.

Low Power In-Memory Implementation of Ternary Neural Networks with Resistive RAM-Based Synapse

Abstract: The design of systems implementing low precision neural networks with emerging memories such as resistive random access memory (RRAM) is a major lead for reducing the energy consumption of artificial intelligence (AI). Multiple works have for example proposed in-memory architectures to implement low power binarized neural networks. These simple neural networks, where synaptic weights and neuronal activations assume binary values, can indeed approach state-of-the-art performance on vision tasks. In this work, we revisit one of these architectures where synapses are implemented in a differential fashion to reduce bit errors, and synaptic weights are read using precharge sense amplifiers. Based on experimental measurements on a hybrid 130 nm CMOS/RRAM chip and on circuit simulation, we show that the same memory array architecture can be used to implement ternary weights instead of binary weights, and that this technique is particularly appropriate if the sense amplifier is operated in near-threshold regime. We also show based on neural network simulation on the CIFAR-10 image recognition task that going from binary to ternary neural networks significantly increases neural network performance. These results highlight that AI circuits function may sometimes be revisited when operated in low power regimes.

Low Power In-Memory Implementation of Ternary Neural Networks with Resistive RAM-Based Synapse

Abstract: The design of systems implementing low precision neural networks with emerging memories such as resistive random access memory (RRAM) is a major lead for reducing the energy consumption of artificial intelligence (AI). Multiple works have for example proposed in-memory architectures to implement low power binarized neural networks. These simple neural networks, where synaptic weights and neuronal activations assume binary values, can indeed approach state-of-the-art performance on vision tasks. In this work, we revisit one of these architectures where synapses are implemented in a differential fashion to reduce bit errors, and synaptic weights are read using precharge sense amplifiers. Based on experimental measurements on a hybrid 130 nm CMOS/RRAM chip and on circuit simulation, we show that the same memory array architecture can be used to implement ternary weights instead of binary weights, and that this technique is particularly appropriate if the sense amplifier is operated in near-threshold regime. We also show based on neural network simulation on the CIFAR-10 image recognition task that going from binary to ternary neural networks significantly increases neural network performance. These results highlight that AI circuits function may sometimes be revisited when operated in low power regimes.

Some Experiments in Test Pattern Generation for FPGA-Implemented Combinational Circuits

Abstract: This paper studies the test pattern generation problem for FPGA implemented combinational circuits. General definitions concerning the specific problem of testing RAM-based FPGAs are first given such as the important concept of manufacturing-oriented test procedure, application-oriented test procedure and AC-non-redundant fault. Then, the test pattern generation problem is discussed and it is pointed out that a high AC-non-redundant fault coverage can be obtained only by using an adequate FPGA representation. It is also shown that test pattern generation performed on the FPGA representation can be significantly accelerated by different techniques. A procedure called TOF is described to validate the proposed approach on benchmark circuits.

A CBRAM-based compact interconnect switch for non-volatile reconfigurable logic circuits

Abstract: This paper presents a 2-to-2 interconnect switch based on Conductive Bridging Random Access Memories (CBRAMs), which can be used to form a switch box in reconfigurable logic circuits like FPGAs. Interconnect switching as well as configuration storage are achieved by the same resistive switching devices. The solution is stable without read disturb and false programming, and brings an area saving of more than two, compared to the current SRAM based circuits. It is a promising breakthrough for including permanent retention mechanisms in embedded systems at low cost.

Emerging memories: resistive switching mechanisms and current status.

Abstract: The resistance switching behaviour of several materials has recently attracted considerable attention for its application in non-volatile memory (NVM) devices, popularly described as resistive random access memories (RRAMs). RRAM is a type of NVM that uses a material(s) that changes the resistance when a voltage is applied. Resistive switching phenomena have been observed in many oxides: (i) binary transition metal oxides (TMOs), e.g. TiO(2), Cr(2)O(3), FeO(x) and NiO; (ii) perovskite-type complex TMOs that are variously functional, paraelectric, ferroelectric, multiferroic and magnetic, e.g. (Ba,Sr)TiO(3), Pb(Zr(x) Ti(1-x))O(3), BiFeO(3) and Pr(x)Ca(1-x)MnO(3); (iii) large band gap high-k dielectrics, e.g. Al(2)O(3) and Gd(2)O(3); (iv) graphene oxides. In the non-oxide category, higher chalcogenides are front runners, e.g. In(2)Se(3) and In(2)Te(3). Hence, the number of materials showing this technologically interesting behaviour for information storage is enormous. Resistive switching in these materials can form the basis for the next generation of NVM, i.e. RRAM, when current semiconductor memory technology reaches its limit in terms of density. RRAMs may be the high-density and low-cost NVMs of the future. A review on this topic is of importance to focus concentration on the most promising materials to accelerate application into the semiconductor industry. This review is a small effort to realize the ambitious goal of RRAMs. Its basic focus is on resistive switching in various materials with particular emphasis on binary TMOs. It also addresses the current understanding of resistive switching behaviour. Moreover, a brief comparison between RRAMs and memristors is included. The review ends with the current status of RRAMs in terms of stability, scalability and switching speed, which are three important aspects of integration onto semiconductors.

Neuromorphic computing using non-volatile memory

Abstract: Dense crossbar arrays of non-volatile memory (NVM) devices represent one possible path for implementing massively-parallel and highly energy-efficient neuromorphic computing systems. We first revie...

Monitoring of environmental exposure to polycyclic aromatic hydrocarbons: a review

Abstract: Polycyclic aromatic hydrocarbons (PAHs) are a large group of organic compounds with two or more fused aromatic rings. They have a relatively low solubility in water, but are highly lipophilic. Most of the PAHs with low vapour pressure in the air are adsorbed on particles. When dissolved in water or adsorbed on particulate matter, PAHs can undergo photodecomposition when exposed to ultraviolet light from solar radiation. In the atmosphere, PAHs can react with pollutants such as ozone, nitrogen oxides and sulfur dioxide, yielding diones, nitro- and dinitro-PAHs, and sulfonic acids, respectively. PAHs may also be degraded by some microorganisms in the soil. PAHs are widespread environmental contaminants resulting from incomplete combustion of organic materials. The occurrence is largely a result of anthropogenic emissions such as fossil fuel-burning, motor vehicle, waste incinerator, oil refining, coke and asphalt production, and aluminum production, etc. PAHs have received increased attention in recent years in air pollution studies because some of these compounds are highly carcinogenic or mutagenic. Eight PAHs (Car-PAHs) typically considered as possible carcinogens are: benzo(a)anthracene, chrysene, benzo(b)fluoranthene, benzo(k)fluoranthene, benzo(a)pyrene (B(a)P), dibenzo(a,h)anthracene, indeno(1,2,3-cd)pyrene and benzo(g,h,i)perylene. In particular, benzo(a)pyrene has been identified as being highly carcinogenic. The US Environmental Protection Agency (EPA) has promulgated 16 unsubstituted PAHs (EPA-PAH) as priority pollutants. Thus, exposure assessments of PAHs in the developing world are important. The scope of this review will be to give an overview of PAH concentrations in various environmental samples and to discuss the advantages and limitations of applying these parameters in the assessment of environmental risks in ecosystems and human health. As it well known, there is an increasing trend to use the behavior of pollutants (i.e. bioaccumulation) as well as pollution-induced biological and biochemical effects on human organisms to evaluate or predict the impact of chemicals on ecosystems. Emphasis in this review will, therefore, be placed on the use of bioaccumulation and biomarker responses in air, soil, water and food, as monitoring tools for the assessment of the risks and hazards of PAH concentrations for the ecosystem, as well as on its limitations.