Roadmap on emerging hardware and technology for machine learning

Abstract

Recent progress in artificial intelligence is largely attributed to the rapid development of machine learning, especially in the algorithm and neural network models. However, it is the performance of the hardware, in particular the energy efficiency of a computing system that sets the fundamental limit of the capability of machine learning. Data-centric computing requires a revolution in hardware systems, since traditional digital computers based on transistors and the von Neumann architecture were not purposely designed for neuromorphic computing. A hardware platform based on emerging devices and new architecture is the hope for future computing with dramatically improved throughput and energy efficiency. Building such a system, nevertheless, faces a number of challenges, ranging from materials selection, device optimization, circuit fabrication, and system integration, to name a few. The aim of this Roadmap is to present a snapshot of emerging hardware technologies that are potentially beneficial for machine learning, providing the Nanotechnology readers with a perspective of challenges and opportunities in this burgeoning field.

Recent Advances in Sequential Infiltration Synthesis (SIS) of Block Copolymers (BCPs)

“…Furthermore, advances in BCPs patterning and SIS techniques can be exploited for the realization of either electrodes and/or active materials of next-generation electronic devices to overcome obstacles of device downscaling and system integration. As an example, BCPs in conjunction with SIS can offer an efficient way for fabricating crossbar arrays of memristive devices for the realization of next-generation computing architectures for neuromorphic-type of data processing, in accordance with the roadmap on emerging hardware and technology for machine learning [ 20 ].…”

“…In addition, the pattern quality of fabricated patterns, in terms of critical dimension and pitch uniformity, was reported to be sufficient for integrated circuit layer manufacturing. The overall lower processing cost and high scalability provided by self-assembly of BCPs could also pave the way for the fabrication of self-assembled crossbar arrays of memristive devices for the realization of next-generation computing architectures, as also underlined in the roadmap on emerging hardware and technology for machine learning [ 20 ]. The great flexibility provided by the BCPs offers the opportunity to employ them as a nanopatterning tool for the design and fabrication of a wide range of functional materials.…”

Adaptive Extreme Edge Computing for Wearable Devices

“…Backpropagation based on online training schemes has also been implemented in several memristive deep learning accelerators (Li et al, 2018a ; Wang et al, 2019d ; Yao et al, 2020 ), showing great success of memristive array on accelerating the deep learning training and adaptive to some device non-ideal characteristics. The readers can refer to more comprehensive review papers for more details (Wang et al, 2020a ; Zhang et al, 2020 ; Berggren et al, 2021 ). In these works, however, the error backpropagation—a backward vector matrix multiplication, and the gradient descent calculation—a vector-vector out-product, are both conducted in hosting computer.…”

Integrated Neuromorphic Photonics: Synapses, Neurons, and Neural Networks

“…[ 36 ] From the pioneering work of utilizing holographic materials for PNN implementation, [ 37 ] neuromorphic photonics are booming, especially in the light of the recent progress of various optical materials [ 38 ] and photonic integrated circuits (PICs). [ 39 ] From the detailed computing performance, a comparison between electronic and PNNs using several experimentally verified photonic components and empirically validated network models, [ 34 ] the bandwidth of PNNs is generally above 100 GHz with the latency below 100 ps, outperforming the electronic ANNs by 2 orders of magnitude, let alone the remarkable energy efficiency on the order of pJ/MAC or even aJ/MAC.…”