Search

Natural intelligence has many dimensions, with some of its most important manifestations being tied to learning about the environment and making behavioral changes. In primates, vision plays a critical role in learning. The underlying biological neural networks contain specialized neurons and synapses which not only sense and process visual stimuli but also learn and adapt with remarkable energy efficiency. Forgetting also plays an active role in learning. Mimicking the adaptive neurobiological mechanisms for seeing, learning, and forgetting can, therefore, accelerate the development of artificial intelligence (AI) and bridge the massive energy gap that exists between AI and biological intelligence. Here, we demonstrate a bioinspired machine vision system based on a 2D phototransistor array fabricated from large-area monolayer molybdenum disulfide (MoS

In spite of recent advancements in artificial neural networks (ANNs), the energy efficiency, multifunctionality, adaptability, and integrated nature of biological neural networks remain largely unimitated by hardware neuromorphic computing systems. Here, we exploit optoelectronic, computing, and programmable memory devices based on emerging two-dimensional (2D) layered materials such as MoS

Graphene has a range of properties that makes it suitable for building devices for the Internet of Things. However, the deployment of such devices will also likely require the development of suitable graphene-based hardware security primitives. Here we report a physically unclonable function (PUF) that exploits disorders in the carrier transport of graphene field-effect transistors. The Dirac voltage, Dirac conductance and carrier mobility values of a large population of graphene field-effect transistors follow Gaussian random distributions, which allow the devices to be used as a PUF. The resulting PUF is resilient to machine learning attacks based on predictive regression models and generative adversarial neural networks. The PUF is also reconfigurable without any physical intervention and/or integration of additional hardware components due to the memristive properties of graphene. Furthermore, we show that the PUF can operate with ultralow power and is scalable, stable over time and reliable against variations in temperature and supply voltage. Disorder in the charge carrier transport of graphene-based field-effect transistors can be used to construct physically unclonable functions that are secure and can withstand advanced computational attacks.

Development of low-power and smart vision sensors is critical for many emerging applications including the acceleration of edge intelligence. In this article, we introduce an active pixel sensor (APS) technology with in-sensor compute capability based on atomically thin two-dimensional (2D) semiconducting material such as monolayer MoS2. The presented 2D APS uses only one programmable phototransistor (1T cell), which significantly reduces the area overhead allowing one to fit 900 pixels in ~0.09 cm2. Phototransistors in the array exploit gate tunable persistent photoconductivity to exhibit high responsivity (~3.6×107 A/W), high specific detectivity (~5.6×1013 Jones), spectral uniformity, and high dynamic range (~80 dB) and electrical programmability to achieve fast reset (~ 100 µs) and in-sensor de-noising capabilities. Commonly encountered problems in the field of 2D material based vision sensors are also resolved by showing near-ideal yield and low device-to-device variation in photoresponse owing to high quality growth, damage-free transfer, and relatively clean fabrication process flow. Remarkably, the energy expenditure by 2D APS was found to be miniscule and in the range of hundreds of femto Joules per pixel. We believe, our low-power 2D APS technology with in-sensor image processing capabilities can be transformative for many edge applications.

The representation of external stimuli in the form of action potentials or spikes constitutes the basis of energy efficient neural computation that emerging spiking neural networks (SNNs) aspire to imitate. With recent evidence suggesting that information in the brain is more often represented by explicit firing times of the neurons rather than mean firing rates, it is imperative to develop novel hardware that can accelerate sparse and spike-timing-based encoding. Here a medium-scale integrated circuit composed of two cascaded three-stage inverters and one XOR logic gate fabricated using a total of 21 memtransistors based on photosensitive 2D monolayer MoS

In-sensor processing, which can reduce the energy and hardware burden for many machine vision applications, is currently lacking in state-of-the-art active pixel sensor (APS) technology. Photosensitive and semiconducting two-dimensional (2D) materials can bridge this technology gap by integrating image capture (sense) and image processing (compute) capabilities in a single device. Here, we introduce a 2D APS technology based on a monolayer MoS

In spite of recent advancements in artificial neural networks (ANNs), the energy efficiency, multifunctionality, adaptability, and integrated nature of the biological neural networks largely remain unimitated in hardware neuromorphic computing systems. Here we exploit optoelectronic, computing, and programmable memory devices based on emerging two-dimensional (2D) layered materials such as MoS2 to demonstrate a monolithically integrated, multi-pixel, and “all-in-one” bio-inspired neural network (BNN) capable of sensing, encoding, learning, forgetting, and inferring at miniscule energy expenditure. We also demonstrate learning adaptability and stimulate learning challenges under specific synaptic conditions to mimic biological learning. Our findings highlight the potential of in-memory computing and sensing based on emerging 2D materials, devices, and integrated circuits not only to overcome the bottleneck of von Neumann computing in conventional CMOS designs but also aid in eliminating peripheral components necessary for competing technologies such as memristors.

Natural intelligence has many dimensions, and in animals, learning about the environment and making behavioral changes are some of its manifestations. In primates vision plays a critical role in learning. The underlying biological neural networks contain specialized neurons and synapses which not only sense and process the visual stimuli but also learns and adapts, with remarkable energy efficiency. Forgetting also plays an active role in learning. Mimicking the adaptive neurobiological mechanisms for seeing, learning, and forgetting can, therefore, accelerate the development of artificial intelligence (AI) and bridge the massive energy gap that exists between AI and biological intelligence. Here we demonstrate a bio-inspired machine vision based on large area grown monolayer 2D phototransistor array integrated with analog, non-volatile, and programmable memory gate-stack that not only enables direct learning, and unsupervised relearning from the visual stimuli but also offers learning adaptability under photopic (bright-light), scotopic (low-light), as well as noisy illumination conditions at miniscule energy expenditure. In short, our “all-in-one” hardware vision platform combines “sensing”, “computing” and “storage” not only to overcome the von Neumann bottleneck of conventional complementary metal oxide semiconductor (CMOS) technology but also to eliminate the need for peripheral circuits and sensors.

In spite of recent advancements in bio-realistic artificial neural networks such as spiking neural networks (SNNs), the energy efficiency, multifunctionality, adaptability, and integrated nature of biological neural networks (BNNs) largely remain unimitated in hardware neuromorphic computing systems. Here we exploit optoelectronic and programmable memory devices based on emerging two-dimensional (2D) layered materials such as MoS2 to demonstrate an “all-in-one” hardware SNN system which is capable of sensing, encoding, unsupervised learning, and inference at miniscule energy expenditure. In short, we have utilized photogating effect in MoS2 based neuromorphic phototransistor for sensing and direct encoding of analog optical information into graded spike trains, we have designed MoS2 based neuromorphic encoding module for conversion of spike trains into spike-count and spike-timing based programming voltages, and finally we have used arrays of programmable MoS2 non-volatile synapses for spike-based unsupervised learning and inference. We also demonstrate adaptability of our SNN for learning under scotopic (low-light) and photopic (bright-light) conditions mimicking neuroplasticity of BNNs. Furthermore, we use our hardware SNN platform to show learning challenges under specific synaptic conditions, which can aid in understanding learning disabilities in BNNs. Our findings highlight the potential of in-memory computing and sensing based on emerging 2D materials, devices, and circuits not only to overcome the bottleneck of von Neumann computing in conventional CMOS designs but also aid in eliminating peripheral components necessary for competing technologies such as memristors, RRAM, PCM, etc. as well as bridge the understanding between neuroscience of learning and machine learning.

Spiking neural networks (SNNs) promise to bridge the gap between artificial neural networks (ANNs) and biological neural networks (BNNs) by exploiting biologically plausible neurons that offer faster inference, lower energy expenditure, and event-driven information processing capabilities. However, implementation of SNNs in future neuromorphic hardware requires hardware encoders analogous to the sensory neurons, which convert external/internal stimulus into spike trains based on specific neural algorithm along with inherent stochasticity. Unfortunately, conventional solid-state transducers are inadequate for this purpose necessitating the development of neural encoders to serve the growing need of neuromorphic computing. Here, we demonstrate a biomimetic device based on a dual gated MoS2 field effect transistor (FET) capable of encoding analog signals into stochastic spike trains following various neural encoding algorithms such as rate-based encoding, spike timing-based encoding, and spike count-based encoding. Two important aspects of neural encoding, namely, dynamic range and encoding precision are also captured in our demonstration. Furthermore, the encoding energy was found to be as frugal as ≈1–5 pJ/spike. Finally, we show fast (≈200 timesteps) encoding of the MNIST data set using our biomimetic device followed by more than 91% accurate inference using a trained SNN., The implementation of spiking neural network in future neuromorphic hardware requires hardware encoder analogous to the sensory neurons. The authors show a biomimetic dual-gated MoS2 field effect transistor capable of encoding analog signals into stochastic spike trains at energy cost of 1–5 pJ/spike.