Neuromorphic architectures for spiking deep neural networks

  title={Neuromorphic architectures for spiking deep neural networks},
  author={G. Indiveri and Federico Corradi and Ning Qiao},
  journal={2015 IEEE International Electron Devices Meeting (IEDM)},
We present a full custom hardware implementation of a deep neural network, built using multiple neuromorphic VLSI devices that integrate analog neuron and synapse circuits together with digital asynchronous logic circuits. The deep network comprises an event-based convolutional stage for feature extraction connected to a spike-based learning stage for feature classification. We describe the properties of the chips used to implement the network and present preliminary experimental results that… 

Figures and Tables from this paper

The Need for Scalable Digital Spiking Neurons
This work proposes a new digital spiking neuron design using spike latency encoding to overcome scalability problems and compares the strengths and the weaknesses in terms of scalability of currently implementeddigital spiking neurons.
Time-Domain Digital-to-Analog Converter for Spiking Neural Network Hardware
A new digital-to-analog converter (DAC) for realizing a synapse circuit of mixed-signal spiking neural networks named "time-domain DAC (TDAC)", which produces weights for converting a digital input code into voltage using one current waveform.
Spiking Neural Networks Hardware Implementations and Challenges
This survey presents the state of the art of hardware implementations of spiking neural networks and the current trends in algorithm elaboration from model selection to training mechanisms and describes the strategies employed to leverage the characteristics of these event-driven algorithms at the hardware level.
Spiking Neural Network with 256 × 256 PCM Array
This work discusses one such neuromorphic chip implementation which has 256 on-chip neuron circuits and 64 k PCM synaptic cells with on- chip learning capability and enables each synaptic cell in the array to operate in asynchronous and parallel fashion, which facilitates scaling up to a larger neural network.
A Training-Efficient Hybrid-Structured Deep Neural Network With Reconfigurable Memristive Synapses
This article designs and fabricates a hybrid-structured DNN (hybrid-DNN), combining both depth-in-space (spatial) and depth- in-time (temporal) deep learning characteristics, and demonstrates its high computing parallelism and energy efficiency with low hardware implementation cost.
A survey on real time processing with spiking neural networks
This survey paper discusses the various hardware implementations of spike neural networks (SNNs) and how they address different issues related to parallel computation of neural network functions.
Analog circuits for mixed-signal neuromorphic computing architectures in 28 nm FD-SOI technology
  • Ning Qiao, G. Indiveri
  • Computer Science
    2017 IEEE SOI-3D-Subthreshold Microelectronics Technology Unified Conference (S3S)
  • 2017
Developing mixed-signal analog-digital neuromorphic circuits in advanced scaled processes poses significant design challenges. We present compact and energy efficient sub-threshold analog synapse and
Benchmarking Physical Performance of Neural Inference Circuits
A consistent and transparent methodology is proposed and used to benchmark this comprehensive set of options across several application cases and identify promising architecture/device combinations.
A versatile neuromorphic system based on simple neuron model
A compact, programmable, versatile, and scalable neuromorphic architecture that effectively alleviates the hardware explosion in fully-connected architecture and effectively realizes different convolution operations which are basic computing operations in convolutional neural networks.
Robust Learning and Recognition of Visual Patterns in Neuromorphic Electronic Agents
A spiking neural network architecture with spike-based learning that enables robust learning and recognition of visual patterns in noisy silicon neural substrate and noisy environments and is used to perform pattern recognition and inference after a training phase with computers and neuromorphic hardware in the loop.


A reconfigurable on-line learning spiking neuromorphic processor comprising 256 neurons and 128K synapses
This paper presents a full-custom mixed-signal VLSI device with neuromorphic learning circuits that emulate the biophysics of real spiking neurons and dynamic synapses for exploring the properties of computational neuroscience models and for building brain-inspired computing systems.
Neuromorphic Electronic Circuits for Building Autonomous Cognitive Systems
This work reviews neuromorphic circuits for emulating neural and synaptic dynamics in real time and discusses the role of biophysically realistic temporal dynamics in hardware neural processing architectures; it is argued how the circuits and networks presented represent a useful set of components for efficiently and elegantly implementing neuromorphic cognition.
A memory-efficient routing method for large-scale spiking neural networks
This work proposes a two stage routing scheme which minimizes the memory requirements needed to implement scalable and reconfigurable spiking neural networks with bounded connectivity and is optimized for the most common and anatomically realistic neural network topologies.
Memory and Information Processing in Neuromorphic Systems
A survey of brain-inspired processor architectures that support models of cortical networks and deep neural networks is presented and the advantages and challenges that need to be addressed for building artificial neural processing systems that can display the richness of behaviors seen in biological systems are presented.
Integration of nanoscale memristor synapses in neuromorphic computing architectures
This paper proposes a novel hybrid memristor-CMOS neuromorphic circuit which represents a radical departure from conventional neuro-computing approaches, as it uses memristors to directly emulate the biophysics and temporal dynamics of real synapses.
Learning Real-World Stimuli in a Neural Network with Spike-Driven Synaptic Dynamics
We present a model of spike-driven synaptic plasticity inspired by experimental observations and motivated by the desire to build an electronic hardware device that can learn to classify complex
Complex Learning in Bio-plausible Memristive Networks
A framework to support complex learning functions by introducing dedicated learning algorithms to a bio-plausible recurrent memristive network with internal dynamics is proposed and the learning of various timing patterns and a complex spatiotemporal pattern of human motor is demonstrated.
An Electronic Synapse Device Based on Metal Oxide Resistive Switching Memory for Neuromorphic Computation
The multilevel capability of metal oxide resistive switching memory was explored for the potential use as a single-element electronic synapse device. TiN/HfOx/AlOx/ Pt resistive switching cells were
Firing patterns in the adaptive exponential integrate-and-fire model
The versatility of a simple two-equation model: the adaptive exponential integrate-and-fire neuron is explored, showing that this model generates multiple firing patterns depending on the choice of parameter values, and a phase diagram describing the transition from one firing type to another is presented.
Adaptive exponential integrate-and-fire model as an effective description of neuronal activity.
The authors' simple model predicts correctly the timing of 96% of the spikes of the detailed model in response to injection of noisy synaptic conductances and has enough expressive power to reproduce qualitatively several electrophysiological classes described in vitro.