This work presents a method for reducing memory consumption to a constant complexity when training deep neural networks. The algorithm is based on the more biologically plausible alternatives of the backpropagation (BP): direct feedback alignment (DFA) and feedback alignment (FA), which use random matrices to propagate error. The proposed method, memory-efficient direct feedback alignment (MEM-DFA), uses higher independence of layers in DFA and allows avoiding storing at once all activation vectors, unlike standard BP, FA, and DFA. Thus, our algorithm's memory usage is constant regardless of the number of layers in a neural network. The method increases the computational cost only by a constant factor of one extra forward pass.
The MEM-DFA, BP, FA, and DFA were evaluated along with their memory profiles on MNIST and CIFAR-10 datasets on various neural network models. Our experiments agree with our theoretical results and show a significant decrease in the memory cost of MEM-DFA compared to the other algorithms.
In this work we present In-Place Activated Batch Normalization (InPlace-ABN) - a novel approach to drastically reduce the training memory footprint of modern deep neural networks in a computationally efficient way. Our solution substitutes the conventionally used succession of BatchNorm + Activation layers with a single plugin layer, hence avoiding invasive framework surgery while providing straightforward applicability for existing deep learning frameworks. We obtain memory savings of up to 50% by dropping intermediate results and by recovering required information during the backward pass th…
Duke University（杜克大学）；University of California, Los Angeles（加州大学洛杉矶分校）
Deep neural networks (DNNs) typically have enough capacity to fit random data by brute force even when conventional data-dependent regularizations focusing on the geometry of the features are imposed. We find out that the reason for this is the inconsistency between the enforced geometry and the standard softmax cross entropy loss. To resolve this, we propose a new framework for data-dependent DNN regularization, the Geometrically-Regularized-Self-Validating neural Networks (GRSVNet). During training, the geometry enforced on one batch of features is simultaneously validated on a separate batc…
Deep neural networks (DNNs) based saliency detection approaches have succeed in recent years, and improved the performance by a great margin via increasingly sophisticated network architecture. Despite the performance improvement, the computational cost is excessively high for such low level visual task. In this work, we propose a light-weighted saliency detection approach with distinctively lower runtime memory cost and model size. We evaluated the performance of our approach on multiple benchmark datasets, and achieved competitive results comparing with state-of-the-art methods on multiple m…
Deep neural networks (DNNs) are computationally/memory-intensive and vulnerable to adversarial attacks, making them prohibitive in some real-world applications. By converting dense models into sparse ones, pruning appears to be a promising solution to reducing the computation/memory cost. This paper studies classification models, especially DNN-based ones, to demonstrate that there exists intrinsic relationships between their sparsity and adversarial robustness. Our analyses reveal, both theoretically and empirically, that nonlinear DNN-based classifiers behave differently under $l_2$ attacks …
Deep Neural Networks (DNNs) have gained immense success in cognitive applications and greatly pushed today's artificial intelligence forward. The biggest challenge in executing DNNs is their extremely data-extensive computations. The computing efficiency in speed and energy is constrained when traditional computing platforms are employed in such computational hungry executions. Spiking neuromorphic computing (SNC) has been widely investigated in deep networks implementation own to their high efficiency in computation and communication. However, weights and signals of DNNs are required to be qu…
Purdue University（普渡大学）；Loyola University Chicago（芝加哥洛约拉大学）
Deep neural networks (DNNs) are successful in many computer vision tasks. However, the most accurate DNNs require millions of parameters and operations, making them energy, computation and memory intensive. This impedes the deployment of large DNNs in low-power devices with limited compute resources. Recent research improves DNN models by reducing the memory requirement, energy consumption, and number of operations without significantly decreasing the accuracy. This paper surveys the progress of low-power deep learning and computer vision, specifically in regards to inference, and discusses th…
Intel Labs（英特尔实验室）；University of Southern California（南加州大学）
Existing deep neural networks (DNNs) that achieve state-of-the-art (SOTA) performance on both clean and adversarially-perturbed images rely on either activation or weight conditioned convolution operations. However, such conditional learning costs additional multiply-accumulate (MAC) or addition operations, increasing inference memory and compute costs. To that end, we present a sparse mixture once for all adversarial training (SMART), that allows a model to train once and then in-situ trade-off between accuracy and robustness, that too at a reduced compute and parameter overhead. In particula…
The use of lower precision has emerged as a popular technique to optimize the compute and storage requirements of complex Deep Neural Networks (DNNs). In the quest for lower precision, recent studies have shown that ternary DNNs (which represent weights and activations by signed ternary values) represent a promising sweet spot, achieving accuracy close to full-precision networks on complex tasks. We propose TiM-DNN, a programmable in-memory accelerator that is specifically designed to execute ternary DNNs. TiM-DNN supports various ternary representations including unweighted {-1,0,1}, symmetri…
Deep Learning is able to solve a plethora of once impossible problems. However, they are vulnerable to input adversarial attacks preventing them from being autonomously deployed in critical applications. Several algorithm-centered works have discussed methods to cause adversarial attacks and improve adversarial robustness of a Deep Neural Network (DNN). In this work, we elicit the advantages and vulnerabilities of hybrid 6T-8T memories to improve the adversarial robustness and cause adversarial attacks on DNNs. We show that bit-error noise in hybrid memories due to erroneous 6T-SRAM cells have…
Deep Neural Networks (DNNs) are computationally and memory intensive, which makes their hardware implementation a challenging task especially for resource constrained devices such as IoT nodes. To address this challenge, this paper introduces a new method to improve DNNs performance by fusing approximate computing with data reuse techniques to be used for image recognition applications. DNNs weights are approximated based on the linear and quadratic approximation methods during the training phase, then, all of the weights are replaced with the linear/quadratic coefficients to execute the infer…
Huazhong University of Science and Technology（华中科技大学）
With pervasive applications of medical imaging in health-care, biomedical image segmentation plays a central role in quantitative analysis, clinical diagno- sis, and medical intervention. Since manual anno- tation su ers limited reproducibility, arduous e orts, and excessive time, automatic segmentation is desired to process increasingly larger scale histopathological data. Recently, deep neural networks (DNNs), par- ticularly fully convolutional networks (FCNs), have been widely applied to biomedical image segmenta- tion, attaining much improved performance. At the same time, quantization of …
Lero, School of Computer Science & Statistics, Trinity College Dublin（都柏林圣三一学院计算机科学与统计学院 Lero）；Synopsys Inc, Dublin（新思科技，都柏林）
Deep neural networks (DNNs) require very large amounts of computation both for training and for inference when deployed in the field. A common approach to implementing DNNs is to recast the most computationally expensive operations as general matrix multiplication (GEMM). However, as we demonstrate in this paper, there are a great many different ways to express DNN convolution operations using GEMM. Although different approaches all perform the same number of operations, the size of temporary data structures differs significantly. Convolution of an input matrix with dimensions $C \times H \tim…
Advanced Research Lab, NavInfo Europe（NavInfo Europe 高级研究实验室）
Deep neural networks (DNNs) fail to learn effectively under label noise and have been shown to memorize random labels which affect their generalization performance. We consider learning in isolation, using one-hot encoded labels as the sole source of supervision, and a lack of regularization to discourage memorization as the major shortcomings of the standard training procedure. Thus, we propose Noisy Concurrent Training (NCT) which leverages collaborative learning to use the consensus between two models as an additional source of supervision. Furthermore, inspired by trial-to-trial variabilit…
University of Chinese Academy of Sciences（中国科学院大学）；Henan University of Technology（河南工业大学）；Ant Financial Services Group（蚂蚁金融服务集团）
Deep neural networks (DNNs) have achieved extraordinary success in numerous areas. However, to attain this success, DNNs often carry a large number of weight parameters, leading to heavy costs of memory and computation resources. Overfitting is also likely to happen in such network when the training data are insufficient. These shortcomings severely hinder the application of DNNs in resource-constrained platforms. In fact, many network weights are known to be redundant and can be removed from the network without much loss of performance. To this end, we introduce a new non-convex integrated tr…
Washington State University（华盛顿州立大学）；University of Iowa（爱荷华大学）；University of Sydney（悉尼大学）
Deep neural networks (DNNs) are becoming increasingly deeper, wider, and non-linear due to the growing demands on prediction accuracy and analysis quality. When training a DNN model, the intermediate activation data must be saved in the memory during forward propagation and then restored for backward propagation. However, state-of-the-art accelerators such as GPUs are only equipped with very limited memory capacities due to hardware design constraints, which significantly limits the maximum batch size and hence performance speedup when training large-scale DNNs. In this paper, we propose a nov…
Deep neural networks (DNNs) have already achieved great success in a lot of application areas and brought profound changes to our society. However, it also raises new security problems, among which how to protect the intellectual property (IP) of DNNs against infringement is one of the most important yet very challenging topics. To deal with this problem, recent studies focus on the IP protection of DNNs by applying digital watermarking, which embeds source information and/or authentication data into DNN models by tuning network parameters directly or indirectly. However, tuning network parame…
Institute of Neuroinformatics（神经信息学研究所）；University of Zurich（苏黎世大学）；ETH Zurich（苏黎世联邦理工学院）
The energy consumed by running large deep neural networks (DNNs) on hardware accelerators is dominated by the need for lots of fast memory to store both states and weights. This large required memory is currently only economically viable through DRAM. Although DRAM is high-throughput and low-cost memory (costing 20X less than SRAM), its long random access latency is bad for the unpredictable access patterns in spiking neural networks (SNNs). In addition, accessing data from DRAM costs orders of magnitude more energy than doing arithmetic with that data. SNNs are energy-efficient if local memor…
Northeastern University（东北大学）；University of Connecticut（康涅狄格大学）
The computing wall and data movement challenges of deep neural networks (DNNs) have exposed the limitations of conventional CMOS-based DNN accelerators. Furthermore, the deep structure and large model size will make DNNs prohibitive to embedded systems and IoT devices, where low power consumption are required. To address these challenges, spin orbit torque magnetic random-access memory (SOT-MRAM) and SOT-MRAM based Processing-In-Memory (PIM) engines have been used to reduce the power consumption of DNNs since SOT-MRAM has the characteristic of near-zero standby power, high density, none-volati…
Massachusetts Institute of Technology（麻省理工学院）；Vanderbilt University（范德堡大学）
Deep neural networks (DNNs) have emerged as key enablers of machine learning. Applying larger DNNs to more diverse applications is an important challenge. The computations performed during DNN training and inference are dominated by operations on the weight matrices describing the DNN. As DNNs incorporate more layers and more neurons per layers, these weight matrices may be required to be sparse because of memory limitations. Sparse DNNs are one possible approach, but the underlying theory is in the early stages of development and presents a number of challenges, including determining the accu…
Interdisciplinary Research Center for Intelligent Secure Systems（智能安全系统跨学科研究中心）
Deep neural networks (DNNs) have demonstrated their effectiveness in a wide range of computer vision tasks, with the state-of-the-art results obtained through complex and deep structures that require intensive computation and memory. Now-a-days, efficient model inference is crucial for consumer applications on resource-constrained platforms. As a result, there is much interest in the research and development of dedicated deep learning (DL) hardware to improve the throughput and energy efficiency of DNNs. Low-precision representation of DNN data-structures through quantization would bring great…
Semi-parametric Memory Consolidation: Towards Brain-like Deep Continual Learning
Centre for Artificial Intelligence and Robotics（人工智能与机器人中心）；Hong Kong Institute of Science and Innovation（香港科学与创新研究院）；Chinese Academy of Sciences（中国科学院）；Department of Computer Science（计算机科学系）；City University of Hong Kong（香港城市大学）；The Department of Neurosurgery（神经外科部门）；Peking University First Hospital（北京大学第一医院）；State Key Laboratory of Multimodal Artificial Intelligence Systems（多模态人工智能系统国家重点实验室）；Institute of Automation（自动化研究所）；Hong Kong SAR, China（香港特别行政区，中国）
The spread of deep learning on embedded devices has prompted the development of numerous methods to optimise the deployment of deep neural networks (DNN). Works have mainly focused on: i) efficient DNN architectures, ii) network optimisation techniques such as pruning and quantisation, iii) optimised algorithms to speed up the execution of the most computational intensive layers and, iv) dedicated hardware to accelerate the data flow and computation. However, there is a lack of research on cross-level optimisation as the space of approaches becomes too large to test and obtain a globally optim…
