One-step regression and classification with cross-point resistive memory arrays

Abstract

One Sentence Summary: Machine learning algorithms such as linear regression and logisticregression are trained in one step with crosspoint resistive memory arrays.2 Abstract: Machine learning has been getting a large attention in the recent years, as a tool to process big data generated by ubiquitous sensors in our daily life. High speed, low energy computing machines are in demand to enable real-time artificial intelligence at the edge, i.e., without the support of a remote frame server in the cloud. Such requirements challenge the complementary metal-oxide-semiconductor (CMOS) technology, which is limited by the Moore's law approaching its end and the communication bottleneck in conventional computing architecture. Novel computing concepts, architectures and devices are thus strongly needed to accelerate data-intensive applications. Here we show a crosspoint resistive memory circuit with feedback configuration can execute linear regression and logistic regression in just one step by computing the pseudoinverse matrix of the data within the memory. The most elementary learning operation, that is the regression of a sequence of data and the classification of a set of data, can thus be executed in one single computational step by the novel technology. One-step learning is further supported by simulations of the prediction of the cost of a house in Boston and the training of a 2-layer neural network for MNIST digit recognition. The results are all obtained in one computational step, thanks to the physical, parallel, and analog computing within the crosspoint array.

Neurohybrid Memristive CMOS-Integrated Systems for Biosensors and Neuroprosthetics

“…Although the local resistive switching effect in memristive devices provides the unique compactness, fast and energyefficient operation of passive memristive arrays (Xia and Yang, 2019), the active arrays integrated with peripheral and control electronics should be always a subject of explicit evaluation and benchmarking depending on the development/prototyping stage (Cai et al, 2019;Zhao et al, 2020). Recently, several reports on such benchmarking have shown potential advantages of memristive chips over conventional ones: 19.7, 6.5 times, and 2 orders of magnitude better energy efficiency compared to the Google's tensor processing unit (TPU), a highly optimized application-specific integrated circuit (ASIC) system, and the state-of-the-art graphics-processing unit (GPU), respectively (Sun et al, 2020;Yao et al, 2020). The performance benchmark of memristive neuromorphic computing system shows 110 times better energy efficiency and 30 times better performance density compared to Tesla V100 GPU.…”

Memristive and CMOS Devices for Neuromorphic Computing

“…Solving the memory bottleneck requires a paradigm shift in architecture, where computation is executed in situ within the data by exploiting, e.g., the ability of memory arrays to implement matrix-vector multiplication (MVM) [10,11]. This novel architectural approach is referred to as in-memory computing, which provides the basis for several outstanding applications, such as pattern classification [12,13], analogue image processing [14], and the solution of linear systems [15,16] and of linear regression problems [17].…”

In‐Memory Hamming Weight Calculation in a 1T1R Memristive Array

“…Now it is widely recognized that this problem could be well addressed by the idea of implementing vector matrix multiplication (VMM) in a memristive crossbar or one‐transistor‐one‐resistor (1T1R) memristive array, [ 6,7 ] which could be embedded as the key block in the mobile and Internet‐of‐things (IoT) hardware. This idea has been demonstrated to accelerate a set of algorithms and solve matrix equations, [ 8–10 ] such as multi‐layer perceptron, [ 11,12 ] CNN, [ 13,14 ] sparse coding, [ 12,15 ] k‐means, [ 16 ] with improved speed and energy efficiency. A very recent work demonstrated in‐memory eigenvector calculation in one step to accelerate the PageRank algorithm for search engines.…”