Hyperdimensional computing with holographic and adaptive encoder.

Introduction: Brain-inspired computing has become an emerging field, where a growing number of works focus on developing algorithms that bring machine learning closer to human brains at the functional level. As one of the promising directions, Hyperdimensional Computing (HDC) is centered around the idea of having holographic and high-dimensional representation as the neural activities in our brains. Such representation is the fundamental enabler for the efficiency and robustness of HDC. However, existing HDC-based algorithms suffer from limitations within the encoder. To some extent, they all rely on manually selected encoders, meaning that the resulting representation is never adapted to the tasks at hand. Methods: In this paper, we propose FLASH, a novel hyperdimensional learning method that incorporates an adaptive and learnable encoder design, aiming at better overall learning performance while maintaining good properties of HDC representation. Current HDC encoders leverage Random Fourier Features (RFF) for kernel correspondence and enable locality-preserving encoding. We propose to learn the encoder matrix distribution via gradient descent and effectively adapt the kernel for a more suitable HDC encoding. Results: Our experiments on various regression datasets show that tuning the HDC encoder can significantly boost the accuracy, surpassing the current HDC-based algorithm and providing faster inference than other baselines, including RFF-based kernel ridge regression. Discussion: The results indicate the importance of an adaptive encoder and customized high-dimensional representation in HDC.

Advancing brain-inspired computing with hybrid neural networks.

Brain-inspired computing, drawing inspiration from the fundamental structure and information-processing mechanisms of the human brain, has gained significant momentum in recent years. It has emerged as a research paradigm centered on brain-computer dual-driven and multi-network integration. One noteworthy instance of this paradigm is the hybrid neural network (HNN), which integrates computer-science-oriented artificial neural networks (ANNs) with neuroscience-oriented spiking neural networks (SNNs). HNNs exhibit distinct advantages in various intelligent tasks, including perception, cognition and learning. This paper presents a comprehensive review of HNNs with an emphasis on their origin, concepts, biological perspective, construction framework and supporting systems. Furthermore, insights and suggestions for potential research directions are provided aiming to propel the advancement of the HNN paradigm.

Multi-scale full spike pattern for semantic segmentation.

Spiking neural networks (SNNs), as the brain-inspired neural networks, encode information in spatio-temporal dynamics. They have the potential to serve as low-power alternatives to artificial neural networks (ANNs) due to their sparse and event-driven nature. However, existing SNN-based models for pixel-level semantic segmentation tasks suffer from poor performance and high memory overhead, failing to fully exploit the computational effectiveness and efficiency of SNNs. To address these challenges, we propose the multi-scale and full spike segmentation network (MFS-Seg), which is based on the deep direct trained SNN and represents the first attempt to train a deep SNN with surrogate gradients for semantic segmentation. Specifically, we design an efficient fully-spike residual block (EFS-Res) to alleviate representation issues caused by spiking noise on different channels. EFS-Res utilizes depthwise separable convolution to improve the distributions of spiking feature maps. The visualization shows that our model can effectively extract the edge features of segmented objects. Furthermore, it can significantly reduce the memory overhead and energy consumption of the network. In addition, we theoretically analyze and prove that EFS-Res can avoid the degradation problem based on block dynamical isometry theory. Experimental results on the Camvid dataset, the DDD17 dataset, and the DSEC-Semantic dataset show that our model achieves comparable performance to the mainstream UNet network with up to 31× fewer parameters, while significantly reducing power consumption by over 13×. Overall, our MFS-Seg model demonstrates promising results in terms of performance, memory efficiency, and energy consumption, showcasing the potential of deep SNNs for semantic segmentation tasks. Our code is available in https://github.com/BICLab/MFS-Seg.

An Improved Dyna-Q Algorithm Inspired by the Forward Prediction Mechanism in the Rat Brain for Mobile Robot Path Planning.

The traditional Model-Based Reinforcement Learning (MBRL) algorithm has high computational cost, poor convergence, and poor performance in robot spatial cognition and navigation tasks, and it cannot fully explain the ability of animals to quickly adapt to environmental changes and learn a variety of complex tasks. Studies have shown that vicarious trial and error (VTE) and the hippocampus forward prediction mechanism in rats and other mammals can be used as key components of action selection in MBRL to support "goal-oriented" behavior. Therefore, we propose an improved Dyna-Q algorithm inspired by the forward prediction mechanism of the hippocampus to solve the above problems and tackle the exploration-exploitation dilemma of Reinforcement Learning (RL). This algorithm alternately presents the potential path in the future for mobile robots and dynamically adjusts the sweep length according to the decision certainty, so as to determine action selection. We test the performance of the algorithm in a two-dimensional maze environment with static and dynamic obstacles, respectively. Compared with classic RL algorithms like State-Action-Reward-State-Action (SARSA) and Dyna-Q, the algorithm can speed up spatial cognition and improve the global search ability of path planning. In addition, our method reflects key features of how the brain organizes MBRL to effectively solve difficult tasks such as navigation, and it provides a new idea for spatial cognitive tasks from a biological perspective.

Exploring neural oscillations during speech perception via surrogate gradient spiking neural networks.

Understanding cognitive processes in the brain demands sophisticated models capable of replicating neural dynamics at large scales. We present a physiologically inspired speech recognition architecture, compatible and scalable with deep learning frameworks, and demonstrate that end-to-end gradient descent training leads to the emergence of neural oscillations in the central spiking neural network. Significant cross-frequency couplings, indicative of these oscillations, are measured within and across network layers during speech processing, whereas no such interactions are observed when handling background noise inputs. Furthermore, our findings highlight the crucial inhibitory role of feedback mechanisms, such as spike frequency adaptation and recurrent connections, in regulating and synchronizing neural activity to improve recognition performance. Overall, on top of developing our understanding of synchronization phenomena notably observed in the human auditory pathway, our architecture exhibits dynamic and efficient information processing, with relevance to neuromorphic technology.

Highly reproducible and CMOS-compatible VO2-based oscillators for brain-inspired computing.

With remarkable electrical and optical switching properties induced at low power and near room temperature (68 °C), vanadium dioxide (VO2) has sparked rising interest in unconventional computing among the phase-change materials research community. The scalability and the potential to compute beyond the von Neumann model make VO2 especially appealing for implementation in oscillating neural networks for artificial intelligence applications, to solve constraint satisfaction problems, and for pattern recognition. Its integration into large networks of oscillators on a Silicon platform still poses challenges associated with the stabilization in the correct oxidation state and the ability to fabricate a structure with predictable electrical behavior showing very low variability. In this work, the role played by the different annealing parameters applied by three methods (slow thermal annealing, flash annealing, and rapid thermal annealing), following the vanadium oxide atomic layer deposition, on the formation of VO2 grains is studied and an optimal substrate stack configuration that minimizes variability between devices is proposed. Material and electrical characterizations are performed on the different films and a step-by-step recipe to build reproducible VO2-based oscillators is presented, which is argued to be made possible thanks to the introduction of a hafnium oxide (HfO2) layer between the silicon substrate and the vanadium oxide layer. Up to seven nearly identical VO2-based devices are contacted simultaneously to create a network of oscillators, paving the way for large-scale implementation of VO2 oscillating neural networks.

Composing recurrent spiking neural networks using locally-recurrent motifs and risk-mitigating architectural optimization.

In neural circuits, recurrent connectivity plays a crucial role in network function and stability. However, existing recurrent spiking neural networks (RSNNs) are often constructed by random connections without optimization. While RSNNs can produce rich dynamics that are critical for memory formation and learning, systemic architectural optimization of RSNNs is still an open challenge. We aim to enable systematic design of large RSNNs via a new scalable RSNN architecture and automated architectural optimization. We compose RSNNs based on a layer architecture called Sparsely-Connected Recurrent Motif Layer (SC-ML) that consists of multiple small recurrent motifs wired together by sparse lateral connections. The small size of the motifs and sparse inter-motif connectivity leads to an RSNN architecture scalable to large network sizes. We further propose a method called Hybrid Risk-Mitigating Architectural Search (HRMAS) to systematically optimize the topology of the proposed recurrent motifs and SC-ML layer architecture. HRMAS is an alternating two-step optimization process by which we mitigate the risk of network instability and performance degradation caused by architectural change by introducing a novel biologically-inspired "self-repairing" mechanism through intrinsic plasticity. The intrinsic plasticity is introduced to the second step of each HRMAS iteration and acts as unsupervised fast self-adaptation to structural and synaptic weight modifications introduced by the first step during the RSNN architectural "evolution." We demonstrate that the proposed automatic architecture optimization leads to significant performance gains over existing manually designed RSNNs: we achieve 96.44% on TI46-Alpha, 94.66% on N-TIDIGITS, 90.28% on DVS-Gesture, and 98.72% on N-MNIST. To the best of the authors' knowledge, this is the first work to perform systematic architecture optimization on RSNNs.

Artificial HfO2/TiOx Synapses with Controllable Memory Window and High Uniformity for Brain-Inspired Computing.

Artificial neural networks, as a game-changer to break up the bottleneck of classical von Neumann architectures, have attracted great interest recently. As a unit of artificial neural networks, memristive devices play a key role due to their similarity to biological synapses in structure, dynamics, and electrical behaviors. To achieve highly accurate neuromorphic computing, memristive devices with a controllable memory window and high uniformity are vitally important. Here, we first report that the controllable memory window of an HfO2/TiOx memristive device can be obtained by tuning the thickness ratio of the sublayer. It was found the memory window increased with decreases in the thickness ratio of HfO2 and TiOx. Notably, the coefficients of variation of the high-resistance state and the low-resistance state of the nanocrystalline HfO2/TiOx memristor were reduced by 74% and 86% compared with the as-deposited HfO2/TiOx memristor. The position of the conductive pathway could be localized by the nanocrystalline HfO2 and TiO2 dot, leading to a substantial improvement in the switching uniformity. The nanocrystalline HfO2/TiOx memristive device showed stable, controllable biological functions, including long-term potentiation, long-term depression, and spike-time-dependent plasticity, as well as the visual learning capability, displaying the great potential application for neuromorphic computing in brain-inspired intelligent systems.

Electrochemical-Memristor-Based Artificial Neurons and Synapses-Fundamentals, Applications, and Challenges.

Artificial neurons and synapses are considered essential for the progress of the future brain-inspired computing, based on beyond von Neumann architectures. Here, a discussion on the common electrochemical fundamentals of biological and artificial cells is provided, focusing on their similarities with the redox-based memristive devices. The driving forces behind the functionalities and the ways to control them by an electrochemical-materials approach are presented. Factors such as the chemical symmetry of the electrodes, doping of the solid electrolyte, concentration gradients, and excess surface energy are discussed as essential to understand, predict, and design artificial neurons and synapses. A variety of two- and three-terminal memristive devices and memristive architectures are presented and their application for solving various problems is shown. The work provides an overview of the current understandings on the complex processes of neural signal generation and transmission in both biological and artificial cells and presents the state-of-the-art applications, including signal transmission between biological and artificial cells. This example is showcasing the possibility for creating bioelectronic interfaces and integrating artificial circuits in biological systems. Prospectives and challenges of the modern technology toward low-power, high-information-density circuits are highlighted.

Sparser spiking activity can be better: Feature Refine-and-Mask spiking neural network for event-based visual recognition.

Event-based visual, a new visual paradigm with bio-inspired dynamic perception and µs level temporal resolution, has prominent advantages in many specific visual scenarios and gained much research interest. Spiking neural network (SNN) is naturally suitable for dealing with event streams due to its temporal information processing capability and event-driven nature. However, existing works SNN neglect the fact that the input event streams are spatially sparse and temporally non-uniform, and just treat these variant inputs equally. This situation interferes with the effectiveness and efficiency of existing SNNs. In this paper, we propose the feature Refine-and-Mask SNN (RM-SNN), which has the ability of self-adaption to regulate the spiking response in a data-dependent way. We use the Refine-and-Mask (RM) module to refine all features and mask the unimportant features to optimize the membrane potential of spiking neurons, which in turn drops the spiking activity. Inspired by the fact that not all events in spatio-temporal streams are task-relevant, we execute the RM module in both temporal and channel dimensions. Extensive experiments on seven event-based benchmarks, DVS128 Gesture, DVS128 Gait, CIFAR10-DVS, N-Caltech101, DailyAction-DVS, UCF101-DVS, and HMDB51-DVS demonstrate that under the multi-scale constraints of input time window, RM-SNN can significantly reduce the network average spiking activity rate while improving the task performance. In addition, by visualizing spiking responses, we analyze why sparser spiking activity can be better. Code.

Efficient human activity recognition with spatio-temporal spiking neural networks.

In this study, we explore Human Activity Recognition (HAR), a task that aims to predict individuals' daily activities utilizing time series data obtained from wearable sensors for health-related applications. Although recent research has predominantly employed end-to-end Artificial Neural Networks (ANNs) for feature extraction and classification in HAR, these approaches impose a substantial computational load on wearable devices and exhibit limitations in temporal feature extraction due to their activation functions. To address these challenges, we propose the application of Spiking Neural Networks (SNNs), an architecture inspired by the characteristics of biological neurons, to HAR tasks. SNNs accumulate input activation as presynaptic potential charges and generate a binary spike upon surpassing a predetermined threshold. This unique property facilitates spatio-temporal feature extraction and confers the advantage of low-power computation attributable to binary spikes. We conduct rigorous experiments on three distinct HAR datasets using SNNs, demonstrating that our approach attains competitive or superior performance relative to ANNs, while concurrently reducing energy consumption by up to 94%.

CBMC: A Biomimetic Approach for Control of a 7-Degree of Freedom Robotic Arm.

Many approaches inspired by brain science have been proposed for robotic control, specifically targeting situations where knowledge of the dynamic model is unavailable. This is crucial because dynamic model inaccuracies and variations can occur during the robot's operation. In this paper, inspired by the central nervous system (CNS), we present a CNS-based Biomimetic Motor Control (CBMC) approach consisting of four modules. The first module consists of a cerebellum-like spiking neural network that employs spiking timing-dependent plasticity to learn the dynamics mechanisms and adjust the synapses connecting the spiking neurons. The second module constructed using an artificial neural network, mimicking the regulation ability of the cerebral cortex to the cerebellum in the CNS, learns by reinforcement learning to supervise the cerebellum module with instructive input. The third and last modules are the cerebral sensory module and the spinal cord module, which deal with sensory input and provide modulation to torque commands, respectively. To validate our method, CBMC was applied to the trajectory tracking control of a 7-DoF robotic arm in simulation. Finally, experiments are conducted on the robotic arm using various payloads, and the results of these experiments clearly demonstrate the effectiveness of the proposed methodology.

Trainable Bilingual Synaptic Functions in Bio-enabled Synaptic Transistors.

The signal transmission of the nervous system is regulated by neurotransmitters. Depending on the type of neurotransmitter released by presynaptic neurons, neuron cells can either be excited or inhibited. Maintaining a balance between excitatory and inhibitory synaptic responses is crucial for the nervous system's versatility, elasticity, and ability to perform parallel computing. On the way to mimic the brain's versatility and plasticity traits, creating a preprogrammed balance between excitatory and inhibitory responses is required. Despite substantial efforts to investigate the balancing of the nervous system, a complex circuit configuration has been suggested to simulate the interaction between excitatory and inhibitory synapses. As a meaningful approach, an optoelectronic synapse for balancing the excitatory and inhibitory responses assisted by light mediation is proposed here by deploying humidity-sensitive chiral nematic phases of known polysaccharide cellulose nanocrystals. The environment-induced pitch tuning changes the polarization of the helicoidal organization, affording different hysteresis effects with the subsequent excitatory and inhibitory nonvolatile behavior in the bio-electrolyte-gated transistors. By applying voltage pulses combined with stimulation of chiral light, the artificial optoelectronic synapse tunes not only synaptic functions but also learning pathways and color recognition. These multifunctional bio-based synaptic field-effect transistors exhibit potential for enhanced parallel neuromorphic computing and robot vision technology.

Auditory perception architecture with spiking neural network and implementation on FPGA.

Spike-based perception brings up a new research idea in the field of neuromorphic engineering. A high-performance biologically inspired flexible spiking neural network (SNN) architecture provides a novel method for the exploration of perception mechanisms and the development of neuromorphic computing systems . In this article, we present a biological-inspired spike-based SNN perception digital system that can realize robust perception. The system employs a fully paralleled pipeline scheme to improve the performance and accelerate the processing of feature extraction. An auditory perception system prototype is realized on ten Intel Cyclone field-programmable gate arrays, which can reach the maximum frequency of 107.28 MHz and the maximum throughput of 5364 Mbps. Our design also achieves the power of 5. 148 W/system and energy efficiency of 845.85 µJ. Our auditory perception implementation is also proved to have superior robustness compared with other SNN systems. We use TIMIT digit speech in noise in accuracy testing. Result shows that it achieves up to 85.75% speech recognition accuracy under obvious noise conditions (signal-to-noise ratio of 20 dB) and maintain small accuracy attenuation with the decline of the signal-to-noise ratio. The overall performance of our proposed system outperforms the state-of-the-art perception system on SNN.

BIDL: a brain-inspired deep learning framework for spatiotemporal processing.

Brain-inspired deep spiking neural network (DSNN) which emulates the function of the biological brain provides an effective approach for event-stream spatiotemporal perception (STP), especially for dynamic vision sensor (DVS) signals. However, there is a lack of generalized learning frameworks that can handle various spatiotemporal modalities beyond event-stream, such as video clips and 3D imaging data. To provide a unified design flow for generalized spatiotemporal processing (STP) and to investigate the capability of lightweight STP processing via brain-inspired neural dynamics, this study introduces a training platform called brain-inspired deep learning (BIDL). This framework constructs deep neural networks, which leverage neural dynamics for processing temporal information and ensures high-accuracy spatial processing via artificial neural network layers. We conducted experiments involving various types of data, including video information processing, DVS information processing, 3D medical imaging classification, and natural language processing. These experiments demonstrate the efficiency of the proposed method. Moreover, as a research framework for researchers in the fields of neuroscience and machine learning, BIDL facilitates the exploration of different neural models and enables global-local co-learning. For easily fitting to neuromorphic chips and GPUs, the framework incorporates several optimizations, including iteration representation, state-aware computational graph, and built-in neural functions. This study presents a user-friendly and efficient DSNN builder for lightweight STP applications and has the potential to drive future advancements in bio-inspired research.

Ion-Movement-Based Synaptic Device for Brain-Inspired Computing.

As the amount of data has grown exponentially with the advent of artificial intelligence and the Internet of Things, computing systems with high energy efficiency, high scalability, and high processing speed are urgently required. Unlike traditional digital computing, which suffers from the von Neumann bottleneck, brain-inspired computing can provide efficient, parallel, and low-power computation based on analog changes in synaptic connections between neurons. Synapse nodes in brain-inspired computing have been typically implemented with dozens of silicon transistors, which is an energy-intensive and non-scalable approach. Ion-movement-based synaptic devices for brain-inspired computing have attracted increasing attention for mimicking the performance of the biological synapse in the human brain due to their low area and low energy costs. This paper discusses the recent development of ion-movement-based synaptic devices for hardware implementation of brain-inspired computing and their principles of operation. From the perspective of the device-level requirements for brain-inspired computing, we address the advantages, challenges, and future prospects associated with different types of ion-movement-based synaptic devices.

Emergence of associative learning in a neuromorphic inference network.

Objective. In the theoretical framework of predictive coding and active inference, the brain can be viewed as instantiating a rich generative model of the world that predicts incoming sensory data while continuously updating its parameters via minimization of prediction errors. While this theory has been successfully applied to cognitive processes-by modelling the activity of functional neural networks at a mesoscopic scale-the validity of the approach when modelling neurons as an ensemble of inferring agents, in a biologically plausible architecture, remained to be explored.Approach.We modelled a simplified cerebellar circuit with individual neurons acting as Bayesian agents to simulate the classical delayed eyeblink conditioning protocol. Neurons and synapses adjusted their activity to minimize their prediction error, which was used as the network cost function. This cerebellar network was then implemented in hardware by replicating digital neuronal elements via a low-power microcontroller.Main results. Persistent changes of synaptic strength-that mirrored neurophysiological observations-emerged via local (neurocentric) prediction error minimization, leading to the expression of associative learning. The same paradigm was effectively emulated in low-power hardware showing remarkably efficient performance compared to conventional neuromorphic architectures.Significance. These findings show that: (a) an ensemble of free energy minimizing neurons-organized in a biological plausible architecture-can recapitulate functional self-organization observed in nature, such as associative plasticity, and (b) a neuromorphic network of inference units can learn unsupervised tasks without embedding predefined learning rules in the circuit, thus providing a potential avenue to a novel form of brain-inspired artificial intelligence.

Artificial Neurons and Synapses Based on Al/a-SiNxOy:H/P+-Si Device with Tunable Resistive Switching from Threshold to Memory.

As the building block of brain-inspired computing, resistive switching memory devices have recently attracted great interest due to their biological function to mimic synapses and neurons, which displays the memory switching or threshold switching characteristic. To make it possible for the Si-based artificial neurons and synapse to be integrated with the neuromorphic chip, the tunable threshold and memory switching characteristic is highly in demand for their perfect compatibility with the mature CMOS technology. We first report artificial neurons and synapses based on the Al/a-SiNxOy:H/P+-Si device with the tunable switching from threshold to memory can be realized by controlling the compliance current. It is found that volatile TS from Al/a-SiNxOy:H/P+-Si device under the lower compliance current is induced by the weak Si dangling bond conductive pathway, which originates from the broken Si-H bonds. While stable nonvolatile MS under the higher compliance current is attributed to the strong Si dangling bond conductive pathway, which is formed by the broken Si-H and Si-O bonds. Theoretical calculation reveals that the conduction mechanism of TS and MS agree with P-F model, space charge limited current model and Ohm's law, respectively. The tunable TS and MS characteristic of Al/a-SiNxOy:H/P+-Si device can be successfully employed to mimic the biological behavior of neurons and synapse including the integrate-and-fire function, paired-pulse facilitation, long-term potentiation and long-term depression as well as spike-timing-dependent plasticity. Our discovery supplies an effective way to construct the neuromorphic devices for brain-inspired computing in the AI period.

GrapHD: Graph-Based Hyperdimensional Memorization for Brain-Like Cognitive Learning.

Memorization is an essential functionality that enables today's machine learning algorithms to provide a high quality of learning and reasoning for each prediction. Memorization gives algorithms prior knowledge to keep the context and define confidence for their decision. Unfortunately, the existing deep learning algorithms have a weak and nontransparent notion of memorization. Brain-inspired HyperDimensional Computing (HDC) is introduced as a model of human memory. Therefore, it mimics several important functionalities of the brain memory by operating with a vector that is computationally tractable and mathematically rigorous in describing human cognition. In this manuscript, we introduce a brain-inspired system that represents HDC memorization capability over a graph of relations. We propose GrapHD, hyperdimensional memorization that represents graph-based information in high-dimensional space. GrapHD defines an encoding method representing complex graph structure while supporting both weighted and unweighted graphs. Our encoder spreads the information of all nodes and edges across into a full holistic representation so that no component is more responsible for storing any piece of information than another. Then, GrapHD defines several important cognitive functionalities over the encoded memory graph. These operations include memory reconstruction, information retrieval, graph matching, and shortest path. Our extensive evaluation shows that GrapHD: (1) significantly enhances learning capability by giving the notion of short/long term memorization to learning algorithms, (2) enables cognitive computing and reasoning over memorization graph, and (3) enables holographic brain-like computation with substantial robustness to noise and failure.

EventHD: Robust and efficient hyperdimensional learning with neuromorphic sensor.

Brain-inspired computing models have shown great potential to outperform today's deep learning solutions in terms of robustness and energy efficiency. Particularly, Hyper-Dimensional Computing (HDC) has shown promising results in enabling efficient and robust cognitive learning. In this study, we exploit HDC as an alternative computational model that mimics important brain functionalities toward high-efficiency and noise-tolerant neuromorphic computing. We present EventHD, an end-to-end learning framework based on HDC for robust, efficient learning from neuromorphic sensors. We first introduce a spatial and temporal encoding scheme to map event-based neuromorphic data into high-dimensional space. Then, we leverage HDC mathematics to support learning and cognitive tasks over encoded data, such as information association and memorization. EventHD also provides a notion of confidence for each prediction, thus enabling self-learning from unlabeled data. We evaluate EventHD efficiency over data collected from Dynamic Vision Sensor (DVS) sensors. Our results indicate that EventHD can provide online learning and cognitive support while operating over raw DVS data without using the costly preprocessing step. In terms of efficiency, EventHD provides 14.2× faster and 19.8× higher energy efficiency than state-of-the-art learning algorithms while improving the computational robustness by 5.9×.