اشترك بالحزمة الذهبية واحصل على وصول غير محدود شمرا أكاديميا
تسجيل مستخدم جديدDynamic Precision Analog Computing for Neural Networks
103
0
0.0
(
0
)
اسأل ChatGPT حول البحث
ﻻ يوجد ملخص باللغة العربية
Analog electronic and optical computing exhibit tremendous advantages over digital computing for accelerating deep learning when operations are executed at low precision. In this work, we derive a relationship between analog precision, which is limited by noise, and digital bit precision. We propose extending analog computing architectures to support varying levels of precision by repeating operations and averaging the result, decreasing the impact of noise. Such architectures enable programmable tradeoffs between precision and other desirable performance metrics such as energy efficiency or throughput. To utilize dynamic precision, we propose a method for learning the precision of each layer of a pre-trained model without retraining network weights. We evaluate this method on analog architectures subject to a variety of noise sources such as shot noise, thermal noise, and weight noise and find that employing dynamic precision reduces energy consumption by up to 89% for computer vision models such as Resnet50 and by 24% for natural language processing models such as BERT. In one example, we apply dynamic precision to a shot-noise limited homodyne optical neural network and simulate inference at an optical energy consumption of 2.7 aJ/MAC for Resnet50 and 1.6 aJ/MAC for BERT with <2% accuracy degradation.
قيم البحث
اقرأ أيضاً
As deep neural networks require tremendous amount of computation and memory, analog computing with emerging memory devices is a promising alternative to digital computing for edge devices. However, because of the increasing simulation time for analog
computing system, it has not been explored. To overcome this issue, analytically approximated simulators are developed, but these models are inaccurate and narrow down the options for peripheral circuits for multiply-accumulate operation (MAC). In this sense, we propose a methodology, SEMULATOR (SiMULATOR by Emulating the analog computing block) which uses a deep neural network to emulate the behavior of crossbar-based analog computing system. With the proposed neural architecture, we experimentally and theoretically shows that it emulates a MAC unit for neural computation. In addition, the simulation time is incomparably reduced when it compared to the circuit simulators such as SPICE.
To rapidly process temporal information at a low metabolic cost, biological neurons integrate inputs as an analog sum but communicate with spikes, binary events in time. Analog neuromorphic hardware uses the same principles to emulate spiking neural
networks with exceptional energy-efficiency. However, instantiating high-performing spiking networks on such hardware remains a significant challenge due to device mismatch and the lack of efficient training algorithms. Here, we introduce a general in-the-loop learning framework based on surrogate gradients that resolves these issues. Using the BrainScaleS-2 neuromorphic system, we show that learning self-corrects for device mismatch resulting in competitive spiking network performance on both vision and speech benchmarks. Our networks display sparse spiking activity with, on average, far less than one spike per hidden neuron and input, perform inference at rates of up to 85 k frames/second, and consume less than 200 mW. In summary, our work sets several new benchmarks for low-energy spiking network processing on analog neuromorphic hardware and paves the way for future on-chip learning algorithms.
Analog photonic solutions offer unique opportunities to address complex computational tasks with unprecedented performance in terms of energy dissipation and speeds, overcoming current limitations of modern computing architectures based on electron f
lows and digital approaches. The lack of modularization and lumped element reconfigurability in photonics has prevented the transition to an all-optical analog computing platform. Here, we explore a nanophotonic platform based on epsilon-near-zero materials capable of solving in the analog domain partial differential equations (PDE). Wavelength stretching in zero-index media enables highly nonlocal interactions within the board based on the conduction of electric displacement, which can be monitored to extract the solution of a broad class of PDE problems. By exploiting control of deposition technique through process parameters, we demonstrate the possibility of implementing the proposed nano-optic processor using CMOS-compatible indium-tin-oxide, whose optical properties can be tuned by carrier injection to obtain programmability at high speeds and low energy requirements. Our nano-optical analog processor can be integrated at chip-scale, processing arbitrary inputs at the speed of light.
Existing graph neural networks (GNNs) largely rely on node embeddings, which represent a node as a vector by its identity, type, or content. However, graphs with unlabeled nodes widely exist in real-world applications (e.g., anonymized social network
s). Previous GNNs either assign random labels to nodes (which introduces artefacts to the GNN) or assign one embedding to all nodes (which fails to distinguish one node from another). In this paper, we analyze the limitation of existing approaches in two types of classification tasks, graph classification and node classification. Inspired by our analysis, we propose two techniques, Dynamic Labeling and Preferential Dynamic Labeling, that satisfy desired properties statistically or asymptotically for each type of the task. Experimental results show that we achieve high performance in various graph-related tasks.
Deep Graph Neural Networks (GNNs) show promising performance on a range of graph tasks, yet at present are costly to run and lack many of the optimisations applied to DNNs. We show, for the first time, how to systematically quantise GNNs with minimal
or no loss in performance using Network Architecture Search (NAS). We define the possible quantisation search space of GNNs. The proposed novel NAS mechanism, named Low Precision Graph NAS (LPGNAS), constrains both architecture and quantisation choices to be differentiable. LPGNAS learns the optimal architecture coupled with the best quantisation strategy for different components in the GNN automatically using back-propagation in a single search round. On eight different datasets, solving the task of classifying unseen nodes in a graph, LPGNAS generates quantised models with significant reductions in both model and buffer sizes but with similar accuracy to manually designed networks and other NAS results. In particular, on the Pubmed dataset, LPGNAS shows a better size-accuracy Pareto frontier compared to seven other manual and searched baselines, offering a 2.3 times reduction in model size but a 0.4% increase in accuracy when compared to the best NAS competitor. Finally, from our collected quantisation statistics on a wide range of datasets, we suggest a W4A8 (4-bit weights, 8-bit activations) quantisation strategy might be the bottleneck for naive GNN quantisations.
الأسئلة المقترحة
سجل دخول لتتمكن من نشر تعليقات
التعليقات
جاري جلب التعليقات
سجل دخول لتتمكن من متابعة معايير البحث التي قمت باختيارها
