arXiv subjects

Naigang Wang

Publications and source records attributed to Naigang Wang.

29 records · Page 2Linked to original sources

arXiv2024

Compressing Recurrent Neural Networks for FPGA-accelerated Implementation in Fluorescence Lifetime Imaging

Fluorescence lifetime imaging (FLI) is an important technique for studying cellular environments and molecular interactions, but its real-time application is limited by slow data acquisition, which requires capturing large time-resolved images and complex post-processing using iterative fitting algorithms. Deep learning (DL) models enable real-time inference, but can be computationally demanding due to complex architectures and large matrix operations. This makes DL models ill-suited for direct implementation on field-programmable gate array (FPGA)-based camera hardware. Model compression is thus crucial for practical deployment for real-time inference generation. In this work, we focus on compressing recurrent neural networks (RNNs), which are well-suited for FLI time-series data processing, to enable deployment on resource-constrained FPGA boards. We perform an empirical evaluation of various compression techniques, including weight reduction, knowledge distillation (KD), post-training quantization (PTQ), and quantization-aware training (QAT), to reduce model size and computational load while preserving inference accuracy. Our compressed RNN model, Seq2SeqLite, achieves a balance between computational efficiency and prediction accuracy, particularly at 8-bit precision. By applying KD, the model parameter size was reduced by 98\% while retaining performance, making it suitable for concurrent real-time FLI analysis on FPGA during data capture. This work represents a big step towards integrating hardware-accelerated real-time FLI analysis for fast biological processes.

eess.IV↗

arXiv2024

Mitigating the Impact of Outlier Channels for Language Model Quantization with Activation Regularization

We consider the problem of accurate quantization for language models, where both the weights and activations are uniformly quantized to 4 bits per parameter, the lowest bitwidth format natively supported by GPU hardware. In this context, the key challenge is activation quantization: it is known that language models contain outlier channels whose values on average are orders of magnitude higher than than other channels, which prevents accurate low-bitwidth quantization with known techniques. We systematically study this phenomena and find that these outlier channels emerge early in training, and that they occur more frequently in layers with residual streams. We then propose a simple strategy which regularizes a layer's inputs via quantization-aware training (QAT) and its outputs via activation kurtosis regularization. We show that regularizing both the inputs and outputs is crucial for preventing a model's "migrating" the difficulty in input quantization to the weights, which makes post-training quantization (PTQ) of weights more difficult. When combined with weight PTQ, we show that our approach can obtain a W4A4 model that performs competitively to the standard-precision W16A16 baseline.

cs.LG↗

arXiv2024

A Provably Effective Method for Pruning Experts in Fine-tuned Sparse Mixture-of-Experts

The sparsely gated mixture of experts (MoE) architecture sends different inputs to different subnetworks, i.e., experts, through trainable routers. MoE reduces the training computation significantly for large models, but its deployment can be still memory or computation expensive for some downstream tasks. Model pruning is a popular approach to reduce inference computation, but its application in MoE architecture is largely unexplored. To the best of our knowledge, this paper provides the first provably efficient technique for pruning experts in finetuned MoE models. We theoretically prove that prioritizing the pruning of the experts with a smaller change of the routers l2 norm from the pretrained model guarantees the preservation of test accuracy, while significantly reducing the model size and the computational requirements. Although our theoretical analysis is centered on binary classification tasks on simplified MoE architecture, our expert pruning method is verified on large vision MoE models such as VMoE and E3MoE finetuned on benchmark datasets such as CIFAR10, CIFAR100, and ImageNet.

cs.LG↗

arXiv2021

Improved Techniques for Quantizing Deep Networks with Adaptive Bit-Widths

Quantizing deep networks with adaptive bit-widths is a promising technique for efficient inference across many devices and resource constraints. In contrast to static methods that repeat the quantization process and train different models for different constraints, adaptive quantization enables us to flexibly adjust the bit-widths of a single deep network during inference for instant adaptation in different scenarios. While existing research shows encouraging results on common image classification benchmarks, this paper investigates how to train such adaptive networks more effectively. Specifically, we present two novel techniques for quantizing deep neural networks with adaptive bit-widths of weights and activations. First, we propose a collaborative strategy to choose a high-precision teacher for transferring knowledge to the low-precision student while jointly optimizing the model with all bit-widths. Second, to effectively transfer knowledge, we develop a dynamic block swapping method by randomly replacing the blocks in the lower-precision student network with the corresponding blocks in the higher-precision teacher network. Extensive experiments on multiple image classification datasets including video classification benchmarks for the first time, well demonstrate the efficacy of our approach over state-of-the-art methods.

cs.CV↗

arXiv2021

4-bit Quantization of LSTM-based Speech Recognition Models

We investigate the impact of aggressive low-precision representations of weights and activations in two families of large LSTM-based architectures for Automatic Speech Recognition (ASR): hybrid Deep Bidirectional LSTM - Hidden Markov Models (DBLSTM-HMMs) and Recurrent Neural Network - Transducers (RNN-Ts). Using a 4-bit integer representation, a naïve quantization approach applied to the LSTM portion of these models results in significant Word Error Rate (WER) degradation. On the other hand, we show that minimal accuracy loss is achievable with an appropriate choice of quantizers and initializations. In particular, we customize quantization schemes depending on the local properties of the network, improving recognition performance while limiting computational time. We demonstrate our solution on the Switchboard (SWB) and CallHome (CH) test sets of the NIST Hub5-2000 evaluation. DBLSTM-HMMs trained with 300 or 2000 hours of SWB data achieves $<$0.5% and $<$1% average WER degradation, respectively. On the more challenging RNN-T models, our quantization strategy limits degradation in 4-bit inference to 1.3%.

cs.CL↗

arXiv2021

ScaleCom: Scalable Sparsified Gradient Compression for Communication-Efficient Distributed Training

Large-scale distributed training of Deep Neural Networks (DNNs) on state-of-the-art platforms is expected to be severely communication constrained. To overcome this limitation, numerous gradient compression techniques have been proposed and have demonstrated high compression ratios. However, most existing methods do not scale well to large scale distributed systems (due to gradient build-up) and/or fail to evaluate model fidelity (test accuracy) on large datasets. To mitigate these issues, we propose a new compression technique, Scalable Sparsified Gradient Compression (ScaleCom), that leverages similarity in the gradient distribution amongst learners to provide significantly improved scalability. Using theoretical analysis, we show that ScaleCom provides favorable convergence guarantees and is compatible with gradient all-reduce techniques. Furthermore, we experimentally demonstrate that ScaleCom has small overheads, directly reduces gradient traffic and provides high compression rates (65-400X) and excellent scalability (up to 64 learners and 8-12X larger batch sizes over standard training) across a wide range of applications (image, language, and speech) without significant accuracy loss.

cs.LG↗

arXiv2021

A Comprehensive Survey on Hardware-Aware Neural Architecture Search

Neural Architecture Search (NAS) methods have been growing in popularity. These techniques have been fundamental to automate and speed up the time consuming and error-prone process of synthesizing novel Deep Learning (DL) architectures. NAS has been extensively studied in the past few years. Arguably their most significant impact has been in image classification and object detection tasks where the state of the art results have been obtained. Despite the significant success achieved to date, applying NAS to real-world problems still poses significant challenges and is not widely practical. In general, the synthesized Convolution Neural Network (CNN) architectures are too complex to be deployed in resource-limited platforms, such as IoT, mobile, and embedded systems. One solution growing in popularity is to use multi-objective optimization algorithms in the NAS search strategy by taking into account execution latency, energy consumption, memory footprint, etc. This kind of NAS, called hardware-aware NAS (HW-NAS), makes searching the most efficient architecture more complicated and opens several questions. In this survey, we provide a detailed review of existing HW-NAS research and categorize them according to four key dimensions: the search space, the search strategy, the acceleration technique, and the hardware cost estimation strategies. We further discuss the challenges and limitations of existing approaches and potential future directions. This is the first survey paper focusing on hardware-aware NAS. We hope it serves as a valuable reference for the various techniques and algorithms discussed and paves the road for future research towards hardware-aware NAS.

cs.LG↗

arXiv2019

Accumulation Bit-Width Scaling For Ultra-Low Precision Training Of Deep Networks

Efforts to reduce the numerical precision of computations in deep learning training have yielded systems that aggressively quantize weights and activations, yet employ wide high-precision accumulators for partial sums in inner-product operations to preserve the quality of convergence. The absence of any framework to analyze the precision requirements of partial sum accumulations results in conservative design choices. This imposes an upper-bound on the reduction of complexity of multiply-accumulate units. We present a statistical approach to analyze the impact of reduced accumulation precision on deep learning training. Observing that a bad choice for accumulation precision results in loss of information that manifests itself as a reduction in variance in an ensemble of partial sums, we derive a set of equations that relate this variance to the length of accumulation and the minimum number of bits needed for accumulation. We apply our analysis to three benchmark networks: CIFAR-10 ResNet 32, ImageNet ResNet 18 and ImageNet AlexNet. In each case, with accumulation precision set in accordance with our proposed equations, the networks successfully converge to the single precision floating-point baseline. We also show that reducing accumulation precision further degrades the quality of the trained network, proving that our equations produce tight bounds. Overall this analysis enables precise tailoring of computation hardware to the application, yielding area- and power-optimal systems.

cs.LG↗

arXiv2018

Training Deep Neural Networks with 8-bit Floating Point Numbers

The state-of-the-art hardware platforms for training Deep Neural Networks (DNNs) are moving from traditional single precision (32-bit) computations towards 16 bits of precision -- in large part due to the high energy efficiency and smaller bit storage associated with using reduced-precision representations. However, unlike inference, training with numbers represented with less than 16 bits has been challenging due to the need to maintain fidelity of the gradient computations during back-propagation. Here we demonstrate, for the first time, the successful training of DNNs using 8-bit floating point numbers while fully maintaining the accuracy on a spectrum of Deep Learning models and datasets. In addition to reducing the data and computation precision to 8 bits, we also successfully reduce the arithmetic precision for additions (used in partial product accumulation and weight updates) from 32 bits to 16 bits through the introduction of a number of key ideas including chunk-based accumulation and floating point stochastic rounding. The use of these novel techniques lays the foundation for a new generation of hardware training platforms with the potential for 2-4x improved throughput over today's systems.

cs.LG↗

arXiv2005

Intermediate ferroelectric orthorhombic and monoclinic MB phases in [110]- electric field cooled Pb(Mg1/3Nb2/3)O3-30%PbTiO3 crystals

Structural phase transformations of [110] electric field cooled Pb(Mg1/3Nb2/3)O3-30%PbTiO3 (PMN-30%PT) crystals have been performed by x-ray diffraction in a field-cooled (FC) condition. A phase sequence of cubic(C)-tetragonal(T)-orthorhombic (O)-monoclinic (MB) was found on field-cooling (FC); and a R-MB-O one was observed with increasing field beginning from the zero field-cooled (ZFC) condition at room temperature. The application of the [110] electric field induced a dramatic change in the phase sequence in the FC condition, compared to the corresponding data for PMN-30%PT crystals in a [001] field, which shows that the phase sequence in the FC condition is altered by the crystallographic direction along which a modest electric field (E) is applied. Only when E is applied along [110] are intermediate O and MB phases observed.

cond-mat.mtrl-sci↗

arXiv2004

X-ray and Neutron Diffraction Investigations of the Structural Phase Transformation Sequence under Electric Field in 0.7Pb(Mg1/3Nb2/3)-0.3PbTiO3 Crystals

The structural phase transformations of 0.7Pb(Mg1/3Nb2/3)O3-0.3PbTiO3 (PMN-30%PT) have been studied using x-ray diffraction (XRD) and neutron scattering as a function of temperature and electric field. We observe the phase transformational sequence (i) cubic (C)-> tetragonal (T)-> rhombohedral (R) in the zero-field-cooled (or ZFC) condition; (ii) C->T-> monoclinic (MC)-> monoclinic (MA) in the field-cooled (or FC) condition; and (iii) R->MA->MC->T with increasing field at fixed temperature beginning from the ZFC condition. Upon removal of the field, the MA phase is stable at room temperature in the FC condition, and also in the ZFC condition with increasing field. Several subtleties of our findings are discussed based on results from thermal expansion and dielectric measurements, including (i) the stability of the MA phase; (ii) a difference in lattice parameters between inside bulk and outside layer regions; and (iii) the diffuse nature of the MA and MC phase transition.

cond-mat.mtrl-sci↗