PyTorchKR Papers

최신 논문

총 4,618편
태그·기간 등 상세 필터

카테고리 탐색

로그인 후 이 태그 구독

arXiv:1908.11263공식 코드 · Papers with Code

PULP-NN: Accelerating Quantized Neural Networks on Parallel Ultra-Low-Power RISC-V Processors

We present PULP-NN, an optimized computing library for a parallel ultra-low-power tightly coupled cluster of RISC-V processors. The key innovation in PULP-NN is a set of kernels for Quantized Neural Network (QNN) inference, targeting byte and sub-byte data types, down to INT-1, tuned for the recent trend toward aggressive quantization in deep neural network inference. The proposed library exploits both the digital signal processing (DSP) extensions available in the PULP RISC-V processors and the cluster's parallelism, achieving up to 15.5 MACs/cycle on INT-8 and improving performance by up to 63x with respect to a sequential implementation on a single RISC-V core implementing the baseline RV32IMC ISA. Using PULP-NN, a CIFAR-10 network on an octa-core cluster runs in 30x and 19.6x less clock cycles than the current state-of-the-art ARM CMSIS-NN library, running on STM32L4 and STM32H7 MCUs, respectively. The proposed library, when running on GAP-8 processor, outperforms by 36.8x and by 7.45x the execution on energy efficient MCUs such as STM32L4 and high-end MCUs such as STM32H7 respectively, when operating at the maximum frequency. The energy efficiency on GAP-8 is 14.1x higher than STM32L4 and 39.5x higher than STM32H7, at the maximum efficiency operating point.

arXiv:1908.11250

Smaller Models, Better Generalization

Reducing network complexity has been a major research focus in recent years with the advent of mobile technology. Convolutional Neural Networks that perform various vision tasks without memory overhaul is the need of the hour. This paper focuses on qualitative and quantitative analysis of reducing the network complexity using an upper bound on the Vapnik-Chervonenkis dimension, pruning, and quantization. We observe a general trend in improvement of accuracies as we quantize the models. We propose a novel loss function that helps in achieving considerable sparsity at comparable accuracies to that of dense models. We compare various regularizations prevalent in the literature and show the superiority of our method in achieving sparser models that generalize well.

arXiv:1908.10396공식 코드 · Papers with Code

Accelerating Large-Scale Inference with Anisotropic Vector Quantization

Quantization based techniques are the current state-of-the-art for scaling maximum inner product search to massive databases. Traditional approaches to quantization aim to minimize the reconstruction error of the database points. Based on the observation that for a given query, the database points that have the largest inner products are more relevant, we develop a family of anisotropic quantization loss functions. Under natural statistical assumptions, we show that quantization with these loss functions leads to a new variant of vector quantization that more greatly penalizes the parallel component of a datapoint's residual relative to its orthogonal component. The proposed approach achieves state-of-the-art results on the public benchmarks available at \url{ann-benchmarks.com}.

arXiv:1908.09756공식 코드 · Papers with Code

Differentiable Product Quantization for End-to-End Embedding Compression

Embedding layers are commonly used to map discrete symbols into continuous embedding vectors that reflect their semantic meanings. Despite their effectiveness, the number of parameters in an embedding layer increases linearly with the number of symbols and poses a critical challenge on memory and storage constraints. In this work, we propose a generic and end-to-end learnable compression framework termed differentiable product quantization (DPQ). We present two instantiations of DPQ that leverage different approximation techniques to enable differentiability in end-to-end learning. Our method can readily serve as a drop-in alternative for any existing embedding layer. Empirically, DPQ offers significant compression ratios (14-238$\times$) at negligible or no performance cost on 10 datasets across three different language tasks.

arXiv:1908.10017

Tiny but Accurate: A Pruned, Quantized and Optimized Memristor Crossbar Framework for Ultra Efficient DNN Implementation

The state-of-art DNN structures involve intensive computation and high memory storage. To mitigate the challenges, the memristor crossbar array has emerged as an intrinsically suitable matrix computation and low-power acceleration framework for DNN applications. However, the high accuracy solution for extreme model compression on memristor crossbar array architecture is still waiting for unraveling. In this paper, we propose a memristor-based DNN framework which combines both structured weight pruning and quantization by incorporating alternating direction method of multipliers (ADMM) algorithm for better pruning and quantization performance. We also discover the non-optimality of the ADMM solution in weight pruning and the unused data path in a structured pruned model. Motivated by these discoveries, we design a software-hardware co-optimization framework which contains the first proposed Network Purification and Unused Path Removal algorithms targeting on post-processing a structured pruned model after ADMM steps. By taking memristor hardware constraints into our whole framework, we achieve extreme high compression ratio on the state-of-art neural network structures with minimum accuracy loss. For quantizing structured pruned model, our framework achieves nearly no accuracy loss after quantizing weights to 8-bit memristor weight representation. We share our models at anonymous link https://bit.ly/2VnMUy0.

arXiv:1908.08898

Incremental Binarization On Recurrent Neural Networks For Single-Channel Source Separation

This paper proposes a Bitwise Gated Recurrent Unit (BGRU) network for the single-channel source separation task. Recurrent Neural Networks (RNN) require several sets of weights within its cells, which significantly increases the computational cost compared to the fully-connected networks. To mitigate this increased computation, we focus on the GRU cells and quantize the feedforward procedure with binarized values and bitwise operations. The BGRU network is trained in two stages. The real-valued weights are pretrained and transferred to the bitwise network, which are then incrementally binarized to minimize the potential loss that can occur from a sudden introduction of quantization. As the proposed binarization technique turns only a few randomly chosen parameters into their binary versions, it gives the network training procedure a chance to gently adapt to the partly quantized version of the network. It eventually achieves the full binarization by incrementally increasing the amount of binarization over the iterations. Our experiments show that the proposed BGRU method produces source separation results greater than that of a real-valued fully connected network, with 11-12 dB mean Signal-to-Distortion Ratio (SDR). A fully binarized BGRU still outperforms a Bitwise Neural Network (BNN) by 1-2 dB even with less number of layers.

arXiv:1908.08200

RATQ: A Universal Fixed-Length Quantizer for Stochastic Optimization

We present Rotated Adaptive Tetra-iterated Quantizer (RATQ), a fixed-length quantizer for gradients in first order stochastic optimization. RATQ is easy to implement and involves only a Hadamard transform computation and adaptive uniform quantization with appropriately chosen dynamic ranges. For noisy gradients with almost surely bounded Euclidean norms, we establish an information theoretic lower bound for optimization accuracy using finite precision gradients and show that RATQ almost attains this lower bound. For mean square bounded noisy gradients, we use a gain-shape quantizer which separately quantizes the Euclidean norm and uses RATQ to quantize the normalized unit norm vector. We establish lower bounds for performance of any optimization procedure and shape quantizer, when used with a uniform gain quantizer. Finally, we propose an adaptive quantizer for gain which when used with RATQ for shape quantizer outperforms uniform gain quantization and is, in fact, close to optimal. As a by-product, we show that our fixed-length quantizer RATQ has almost the same performance as the optimal variable-length quantizers for distributed mean estimation. Also, we obtain an efficient quantizer for Gaussian vectors which attains a rate very close to the Gaussian rate-distortion function and is, in fact, universal for subgaussian input vectors.

arXiv:1908.06845

Deep Task-Based Quantization

Quantizers play a critical role in digital signal processing systems. Recent works have shown that the performance of quantization systems acquiring multiple analog signals using scalar analog-to-digital converters (ADCs) can be significantly improved by properly processing the analog signals prior to quantization. However, the design of such hybrid quantizers is quite complex, and their implementation requires complete knowledge of the statistical model of the analog signal, which may not be available in practice. In this work we design data-driven task-oriented quantization systems with scalar ADCs, which determine how to map an analog signal into its digital representation using deep learning tools. These representations are designed to facilitate the task of recovering underlying information from the quantized signals, which can be a set of parameters to estimate, or alternatively, a classification task. By utilizing deep learning, we circumvent the need to explicitly recover the system model and to find the proper quantization rule for it. Our main target application is multiple-input multiple-output (MIMO) communication receivers, which simultaneously acquire a set of analog signals, and are commonly subject to constraints on the number of bits. Our results indicate that, in a MIMO channel estimation setup, the proposed deep task-bask quantizer is capable of approaching the optimal performance limits dictated by indirect rate-distortion theory, achievable using vector quantizers and requiring complete knowledge of the underlying statistical model. Furthermore, for a symbol detection scenario, it is demonstrated that the proposed approach can realize reliable bit-efficient hybrid MIMO receivers capable of setting their quantization rule in light of the task, e.g., to minimize the bit error rate.

arXiv:1908.06077공식 코드 · Papers with Code

NUQSGD: Improved Communication Efficiency for Data-parallel SGD via Nonuniform Quantization

As the size and complexity of models and datasets grow, so does the need for communication-efficient variants of stochastic gradient descent that can be deployed on clusters to perform model fitting in parallel. Alistarh et al. (2017) describe two variants of data-parallel SGD that quantize and encode gradients to lessen communication costs. For the first variant, QSGD, they provide strong theoretical guarantees. For the second variant, which we call QSGDinf, they demonstrate impressive empirical gains for distributed training of large neural networks. Building on their work, we propose an alternative scheme for quantizing gradients and show that it yields stronger theoretical guarantees than exist for QSGD while matching the empirical performance of QSGDinf.

arXiv:1908.05033

Differentiable Soft Quantization: Bridging Full-Precision and Low-Bit Neural Networks

Hardware-friendly network quantization (e.g., binary/uniform quantization) can efficiently accelerate the inference and meanwhile reduce memory consumption of the deep neural networks, which is crucial for model deployment on resource-limited devices like mobile phones. However, due to the discreteness of low-bit quantization, existing quantization methods often face the unstable training process and severe performance degradation. To address this problem, in this paper we propose Differentiable Soft Quantization (DSQ) to bridge the gap between the full-precision and low-bit networks. DSQ can automatically evolve during training to gradually approximate the standard quantization. Owing to its differentiable property, DSQ can help pursue the accurate gradients in backward propagation, and reduce the quantization loss in forward process with an appropriate clipping range. Extensive experiments over several popular network structures show that training low-bit neural networks with DSQ can consistently outperform state-of-the-art quantization methods. Besides, our first efficient implementation for deploying 2 to 4-bit DSQ on devices with ARM architecture achieves up to 1.7$\times$ speed up, compared with the open-source 8-bit high-performance inference framework NCNN. [31]

arXiv:1908.04685

Learn to Compress CSI and Allocate Resources in Vehicular Networks

Resource allocation has a direct and profound impact on the performance of vehicle-to-everything (V2X) networks. In this paper, we develop a hybrid architecture consisting of centralized decision making and distributed resource sharing (the C-Decision scheme) to maximize the long-term sum rate of all vehicles. To reduce the network signaling overhead, each vehicle uses a deep neural network to compress its observed information that is thereafter fed back to the centralized decision making unit. The centralized decision unit employs a deep Q-network to allocate resources and then sends the decision results to all vehicles. We further adopt a quantization layer for each vehicle that learns to quantize the continuous feedback. In addition, we devise a mechanism to balance the transmission of vehicle-to-vehicle (V2V) links and vehicle-to-infrastructure (V2I) links. To further facilitate distributed spectrum sharing, we also propose a distributed decision making and spectrum sharing architecture (the D-Decision scheme) for each V2V link. Through extensive simulation results, we demonstrate that the proposed C-Decision and D-Decision schemes can both achieve near-optimal performance and are robust to feedback interval variations, input noise, and feedback noise.

arXiv:1908.04680

Effective Training of Convolutional Neural Networks with Low-bitwidth Weights and Activations

This paper tackles the problem of training a deep convolutional neural network of both low-bitwidth weights and activations. Optimizing a low-precision network is very challenging due to the non-differentiability of the quantizer, which may result in substantial accuracy loss. To address this, we propose three practical approaches, including (i) progressive quantization; (ii) stochastic precision; and (iii) joint knowledge distillation to improve the network training. First, for progressive quantization, we propose two schemes to progressively find good local minima. Specifically, we propose to first optimize a net with quantized weights and subsequently quantize activations. This is in contrast to the traditional methods which optimize them simultaneously. Furthermore, we propose a second progressive quantization scheme which gradually decreases the bit-width from high-precision to low-precision during training. Second, to alleviate the excessive training burden due to the multi-round training stages, we further propose a one-stage stochastic precision strategy to randomly sample and quantize sub-networks while keeping other parts in full-precision. Finally, we adopt a novel learning scheme to jointly train a full-precision model alongside the low-precision one. By doing so, the full-precision model provides hints to guide the low-precision model training and significantly improves the performance of the low-precision network. Extensive experiments on various datasets (e.g., CIFAR-100, ImageNet) show the effectiveness of the proposed methods.

arXiv:1908.03883공식 코드 · Papers with Code

Unsupervised Neural Quantization for Compressed-Domain Similarity Search

We tackle the problem of unsupervised visual descriptors compression, which is a key ingredient of large-scale image retrieval systems. While the deep learning machinery has benefited literally all computer vision pipelines, the existing state-of-the-art compression methods employ shallow architectures, and we aim to close this gap by our paper. In more detail, we introduce a DNN architecture for the unsupervised compressed-domain retrieval, based on multi-codebook quantization. The proposed architecture is designed to incorporate both fast data encoding and efficient distances computation via lookup tables. We demonstrate the exceptional advantage of our scheme over existing quantization approaches on several datasets of visual descriptors via outperforming the previous state-of-the-art by a large margin.

arXiv:1908.03447

Learn to Allocate Resources in Vehicular Networks

Resource allocation has a direct and profound impact on the performance of vehicle-to-everything (V2X) networks. Considering the dynamic nature of vehicular environments, it is appealing to devise a decentralized strategy to perform effective resource sharing. In this paper, we exploit deep learning to promote coordination among multiple vehicles and propose a hybrid architecture consisting of centralized decision making and distributed resource sharing to maximize the long-term sum rate of all vehicles. To reduce the network signaling overhead, each vehicle uses a deep neural network to compress its own observed information that is thereafter fed back to the centralized decision-making unit, which employs a deep Q-network to allocate resources and then sends the decision results to all vehicles. We further adopt a quantization layer for each vehicle that learns to quantize the continuous feedback. Extensive simulation results demonstrate that the proposed hybrid architecture can achieve near-optimal performance. Meanwhile, there exists an optimal number of continuous feedback and binary feedback, respectively. Besides, this architecture is robust to different feedback intervals, input noise, and feedback noise.

arXiv:1908.02386

Cheetah: Mixed Low-Precision Hardware & Software Co-Design Framework for DNNs on the Edge

Low-precision DNNs have been extensively explored in order to reduce the size of DNN models for edge devices. Recently, the posit numerical format has shown promise for DNN data representation and compute with ultra-low precision in [5..8]-bits. However, previous studies were limited to studying posit for DNN inference only. In this paper, we propose the Cheetah framework, which supports both DNN training and inference using posits, as well as other commonly used formats. Additionally, the framework is amenable for different quantization approaches and supports mixed-precision floating point and fixed-point numerical formats. Cheetah is evaluated on three datasets: MNIST, Fashion MNIST, and CIFAR-10. Results indicate that 16-bit posits outperform 16-bit floating point in DNN training. Furthermore, performing inference with [5..8]-bit posits improves the trade-off between performance and energy-delay-product over both [5..8]-bit float and fixed-point.

arXiv:1908.01073공식 코드 · Papers with Code

U-Net Fixed-Point Quantization for Medical Image Segmentation

Model quantization is leveraged to reduce the memory consumption and the computation time of deep neural networks. This is achieved by representing weights and activations with a lower bit resolution when compared to their high precision floating point counterparts. The suitable level of quantization is directly related to the model performance. Lowering the quantization precision (e.g. 2 bits), reduces the amount of memory required to store model parameters and the amount of logic required to implement computational blocks, which contributes to reducing the power consumption of the entire system. These benefits typically come at the cost of reduced accuracy. The main challenge is to quantize a network as much as possible, while maintaining the performance accuracy. In this work, we present a quantization method for the U-Net architecture, a popular model in medical image segmentation. We then apply our quantization algorithm to three datasets: (1) the Spinal Cord Gray Matter Segmentation (GM), (2) the ISBI challenge for segmentation of neuronal structures in Electron Microscopic (EM), and (3) the public National Institute of Health (NIH) dataset for pancreas segmentation in abdominal CT scans. The reported results demonstrate that with only 4 bits for weights and 6 bits for activations, we obtain 8 fold reduction in memory requirements while loosing only 2.21%, 0.57% and 2.09% dice overlap score for EM, GM and NIH datasets respectively. Our fixed point quantization provides a flexible trade off between accuracy and memory requirement which is not provided by previous quantization methods for U-Net such as TernaryNet.

arXiv:1908.01477

GDRQ: Group-based Distribution Reshaping for Quantization

Low-bit quantization is challenging to maintain high performance with limited model capacity (e.g., 4-bit for both weights and activations). Naturally, the distribution of both weights and activations in deep neural network are Gaussian-like. Nevertheless, due to the limited bitwidth of low-bit model, uniform-like distributed weights and activations have been proved to be more friendly to quantization while preserving accuracy~\cite{Han2015Learning}. Motivated by this, we propose Scale-Clip, a Distribution Reshaping technique that can reshape weights or activations into a uniform-like distribution in a dynamic manner. Furthermore, to increase the model capability for a low-bit model, a novel Group-based Quantization algorithm is proposed to split the filters into several groups. Different groups can learn different quantization parameters, which can be elegantly merged in to batch normalization layer without extra computational cost in the inference stage. Finally, we integrate Scale-Clip technique with Group-based Quantization algorithm and propose the Group-based Distribution Reshaping Quantization (GDQR) framework to further improve the quantization performance. Experiments on various networks (e.g. VGGNet and ResNet) and vision tasks (e.g. classification, detection and segmentation) demonstrate that our framework achieves good performance.

arXiv:1908.00735공식 코드 · Papers with Code

Efficient computation of counterfactual explanations of LVQ models

The increasing use of machine learning in practice and legal regulations like EU's GDPR cause the necessity to be able to explain the prediction and behavior of machine learning models. A prominent example of particularly intuitive explanations of AI models in the context of decision making are counterfactual explanations. Yet, it is still an open research problem how to efficiently compute counterfactual explanations for many models. We investigate how to efficiently compute counterfactual explanations for an important class of models, prototype-based classifiers such as learning vector quantization models. In particular, we derive specific convex and non-convex programs depending on the used metric.

arXiv:1908.00347공식 코드 · Papers with Code

Central Similarity Quantization for Efficient Image and Video Retrieval

Existing data-dependent hashing methods usually learn hash functions from pairwise or triplet data relationships, which only capture the data similarity locally, and often suffer from low learning efficiency and low collision rate. In this work, we propose a new \emph{global} similarity metric, termed as \emph{central similarity}, with which the hash codes of similar data pairs are encouraged to approach a common center and those for dissimilar pairs to converge to different centers, to improve hash learning efficiency and retrieval accuracy. We principally formulate the computation of the proposed central similarity metric by introducing a new concept, i.e., \emph{hash center} that refers to a set of data points scattered in the Hamming space with a sufficient mutual distance between each other. We then provide an efficient method to construct well separated hash centers by leveraging the Hadamard matrix and Bernoulli distributions. Finally, we propose the Central Similarity Quantization (CSQ) that optimizes the central similarity between data points w.r.t.\ their hash centers instead of optimizing the local similarity. CSQ is generic and applicable to both image and video hashing scenarios. Extensive experiments on large-scale image and video retrieval tasks demonstrate that CSQ can generate cohesive hash codes for similar data pairs and dispersed hash codes for dissimilar pairs, achieving a noticeable boost in retrieval performance, i.e. 3\%-20\% in mAP over the previous state-of-the-arts. The code is at: \url{https://github.com/yuanli2333/Hadamard-Matrix-for-hashing}

arXiv:1907.11900공식 코드 · Papers with Code

DeepCABAC: A Universal Compression Algorithm for Deep Neural Networks

The field of video compression has developed some of the most sophisticated and efficient compression algorithms known in the literature, enabling very high compressibility for little loss of information. Whilst some of these techniques are domain specific, many of their underlying principles are universal in that they can be adapted and applied for compressing different types of data. In this work we present DeepCABAC, a compression algorithm for deep neural networks that is based on one of the state-of-the-art video coding techniques. Concretely, it applies a Context-based Adaptive Binary Arithmetic Coder (CABAC) to the network's parameters, which was originally designed for the H.264/AVC video coding standard and became the state-of-the-art for lossless compression. Moreover, DeepCABAC employs a novel quantization scheme that minimizes the rate-distortion function while simultaneously taking the impact of quantization onto the accuracy of the network into account. Experimental results show that DeepCABAC consistently attains higher compression rates than previously proposed coding techniques for neural network compression. For instance, it is able to compress the VGG16 ImageNet model by x63.6 with no loss of accuracy, thus being able to represent the entire network with merely 8.7MB. The source code for encoding and decoding can be found at https://github.com/fraunhoferhhi/DeepCABAC.