Deep Neural Networks (DNNs) are known to be vulnerable to adversarial examples. Further, these adversarial examples are found to be transferable from the source network in which they are crafted to a black-box target network. As the trend of using deep learning on embedded devices grows, it becomes relevant to study the transferability properties of adversarial examples among compressed networks. In this paper, we consider quantization as a network compression technique and evaluate the performance of transfer-based attacks when the source and target networks are quantized at different bitwidths. We explore how algorithm specific properties affect transferability by considering various adversarial example generation algorithms. Furthermore, we examine transferability in a more realistic scenario where the source and target networks may differ in bitwidth and other model-related properties like capacity and architecture. We find that although quantization reduces transferability, certain attack types demonstrate an ability to enhance it. Additionally, the average transferability of adversarial examples among quantized versions of a network can be used to estimate the transferability to quantized target networks with varying capacity and architecture.
In frequency-division duplexing (FDD) multiple-input multiple-output (MIMO) systems, obtaining accurate downlink channel state information (CSI) for precoding is vastly challenging due to the tremendous feedback overhead with the growing number of antennas. Utilizing uplink pilots for downlink CSI estimation is a promising approach that can eliminate CSI feedback. However, the downlink CSI estimation accuracy diminishes significantly as the number of channel paths increases, resulting in reduced spectral efficiency. In this paper, we demonstrate that achieving downlink spectral efficiency comparable to perfect CSI is feasible by combining uplink CSI with limited downlink CSI feedback information. Our proposed downlink CSI feedback strategy transmits quantized phase information of downlink channel paths, deviating from conventional limited methods. We put forth a mean square error (MSE)-optimal downlink channel reconstruction method by jointly exploiting the uplink CSI and the limited downlink CSI. Armed with the MSE-optimal estimator, we derive the MSE as a function of the number of feedback bits for phase quantization. Subsequently, we present an optimal feedback bit allocation method for minimizing the MSE in the reconstructed channel through phase quantization. Utilizing a robust downlink precoding technique, we establish that the proposed downlink channel reconstruction method is sufficient for attaining a sum-spectral efficiency comparable to perfect CSI.
Conventional data compression schemes aim at implementing a trade-off between the rate required to represent the compressed data and the resulting distortion between the original and reconstructed data. However, in more and more applications, what is desired is not reconstruction accuracy but the quality of the realization of a certain task by the receiver. In this paper, the receiver task is modeled by an optimization problem whose parameters have to be compressed by the transmitter. Motivated by applications such as the smart grid, this paper focuses on a goal function which is of $L_p$-norm-type. The aim is to design the precoding, quantization, and decoding stages such that the maximum of the goal function obtained with the compressed version of the parameters is as close as possible to the maximum obtained without compression. The numerical analysis, based on real smart grid signals, clearly shows the benefits of the proposed approach compared to the conventional distortion-based compression paradigm.
In this paper, we propose feature-based federated transfer learning as a novel approach to improve communication efficiency by reducing the uplink payload by multiple orders of magnitude compared to that of existing approaches in federated learning and federated transfer learning. Specifically, in the proposed feature-based federated learning, we design the extracted features and outputs to be uploaded instead of parameter updates. For this distributed learning model, we determine the required payload and provide comparisons with the existing schemes. Subsequently, we analyze the robustness of feature-based federated transfer learning against packet loss, data insufficiency, and quantization. Finally, we address privacy considerations by defining and analyzing label privacy leakage and feature privacy leakage, and investigating mitigating approaches. For all aforementioned analyses, we evaluate the performance of the proposed learning scheme via experiments on an image classification task and a natural language processing task to demonstrate its effectiveness.
Neural audio coding has emerged as a vivid research direction by promising good audio quality at very low bitrates unachievable by classical coding techniques. Here, end-to-end trainable autoencoder-like models represent the state of the art, where a discrete representation in the bottleneck of the autoencoder is learned. This allows for efficient transmission of the input audio signal. The learned discrete representation of neural codecs is typically generated by applying a quantizer to the output of the neural encoder. In almost all state-of-the-art neural audio coding approaches, this quantizer is realized as a Vector Quantizer (VQ) and a lot of effort has been spent to alleviate drawbacks of this quantization technique when used together with a neural audio coder. In this paper, we propose and analyze simple alternatives to VQ, which are based on projected Scalar Quantization (SQ). These quantization techniques do not need any additional losses, scheduling parameters or codebook storage thereby simplifying the training of neural audio codecs. For real-time speech communication applications, these neural codecs are required to operate at low complexity, low latency and at low bitrates. We address those challenges by proposing a new causal network architecture that is based on SQ and a Short-Time Fourier Transform (STFT) representation. The proposed method performs particularly well in the very low complexity and low bitrate regime.
Generative Artificial Intelligence (GAI) is taking the world by storm with its unparalleled content creation ability. Large Language Models (LLMs) are at the forefront of this movement. However, the significant resource demands of LLMs often require cloud hosting, which raises issues regarding privacy, latency, and usage limitations. Although edge intelligence has long been utilized to solve these challenges by enabling real-time AI computation on ubiquitous edge resources close to data sources, most research has focused on traditional AI models and has left a gap in addressing the unique characteristics of LLM inference, such as considerable model size, auto-regressive processes, and self-attention mechanisms. In this paper, we present an edge intelligence optimization problem tailored for LLM inference. Specifically, with the deployment of the batching technique and model quantization on resource-limited edge devices, we formulate an inference model for transformer decoder-based LLMs. Furthermore, our approach aims to maximize the inference throughput via batch scheduling and joint allocation of communication and computation resources, while also considering edge resource constraints and varying user requirements of latency and accuracy. To address this NP-hard problem, we develop an optimal Depth-First Tree-Searching algorithm with online tree-Pruning (DFTSP) that operates within a feasible time complexity. Simulation results indicate that DFTSP surpasses other batching benchmarks in throughput across diverse user settings and quantization techniques, and it reduces time complexity by over 45% compared to the brute-force searching method.
Large Language Models (LLMs) have distinguished themselves with outstanding performance in complex language modeling tasks, yet they come with significant computational and storage challenges. This paper explores the potential of quantization to mitigate these challenges. We systematically study the combined application of three well-known post-training techniques, SmoothQuant, AWQ, and GPTQ, and provide a comprehensive analysis of their interactions and implications for advancing LLM quantization. We enhance the versatility of these methods by enabling quantization to microscaling (MX) formats, extending the applicability of these PTQ algorithms beyond their original fixed-point format targets. We show that combining different PTQ methods enables us to quantize models to 4-bit weights and 8-bit activations using the MXINT format with negligible accuracy loss compared to the uncompressed baseline.
Large language models (LLMs) can now handle longer sequences of tokens, enabling complex tasks like book understanding and generating lengthy novels. However, the key-value (KV) cache required for LLMs consumes substantial memory as context length increasing, becoming the bottleneck for deployment. In this paper, we present a strategy called SKVQ, which stands for sliding-window KV cache quantization, to address the issue of extremely low bitwidth KV cache quantization. To achieve this, SKVQ rearranges the channels of the KV cache in order to improve the similarity of channels in quantization groups, and applies clipped dynamic quantization at the group level. Additionally, SKVQ ensures that the most recent window tokens in the KV cache are preserved with high precision. This helps maintain the accuracy of a small but important portion of the KV cache.SKVQ achieves high compression ratios while maintaining accuracy. Our evaluation on LLMs demonstrates that SKVQ surpasses previous quantization approaches, allowing for quantization of the KV cache to 2-bit keys and 1.5-bit values with minimal loss of accuracy. With SKVQ, it is possible to process context lengths of up to 1M on an 80GB memory GPU for a 7b model and up to 7 times faster decoding.
Recent large language models (LLMs) employ billions of parameters to enable broad problem-solving capabilities. Such language models also tend to be memory-bound because of the dominance of matrix-vector and matrix-matrix multiplications with low arithmetic intensity. Therefore, optimizing the memory footprint and traffic is an important optimization direction for LLMs today. Model compression methods such as quantization and parameter pruning have been actively explored to achieve memory footprint and traffic optimization. However, the accuracy-efficiency trade-off of rank pruning (i.e., low-rank decomposition) for LLMs is not well-understood yet. Therefore, in this work, we characterize the accuracy-efficiency trade-off of a low-rank decomposition method, specifically Tucker decomposition, on recent language models, including an open-source LLM, Llama 2. We formalize the low-rank decomposition design space and show that the decomposition design space is enormous (e.g., O($2^{39}$) for Llama2-7B). To navigate such a vast design space, we formulate it and perform thorough case studies of accuracy-efficiency trade-offs using six widely used LLM benchmarks on BERT and Llama 2 models. Our results show that we can achieve a 9\% model size reduction with minimal accuracy drops, which range from 4\%p (\%p refers to "percentage point," which refers to the absolute difference between two percentage numbers; 74\% -> 78\% = 4\%p increase) to 10\%p, depending on the difficulty of the benchmark, without any retraining to recover accuracy after decomposition. The results show that low-rank decomposition can be a promising direction for LLM-based applications that require real-time service at scale (e.g., AI agent and real-time coding assistant), where the latency is as important as the model accuracy.
This paper focuses on the task of quality enhancement for compressed videos. Although deep network-based video restorers achieve impressive progress, most of the existing methods lack a structured design to optimally leverage the priors within compression codecs. Since the quality degradation of the video is primarily induced by the compression algorithm, a new paradigm is urgently needed for a more ``conscious'' process of quality enhancement. As a result, we propose the Compression-Realized Deep Structural Network (CRDS), introducing three inductive biases aligned with the three primary processes in the classic compression codec, merging the strengths of classical encoder architecture with deep network capabilities. Inspired by the residual extraction and domain transformation process in the codec, a pre-trained Latent Degradation Residual Auto-Encoder is proposed to transform video frames into a latent feature space, and the mutual neighborhood attention mechanism is integrated for precise motion estimation and residual extraction. Furthermore, drawing inspiration from the quantization noise distribution of the codec, CRDS proposes a novel Progressive Denoising framework with intermediate supervision that decomposes the quality enhancement into a series of simpler denoising sub-tasks. Experimental results on datasets like LDV 2.0 and MFQE 2.0 indicate our approach surpasses state-of-the-art models.
Lane detection (LD) plays a crucial role in enhancing the L2+ capabilities of autonomous driving, capturing widespread attention. The Post-Processing Quantization (PTQ) could facilitate the practical application of LD models, enabling fast speeds and limited memories without labeled data. However, prior PTQ methods do not consider the complex LD outputs that contain physical semantics, such as offsets, locations, etc., and thus cannot be directly applied to LD models. In this paper, we pioneeringly investigate semantic sensitivity to post-processing for lane detection with a novel Lane Distortion Score. Moreover, we identify two main factors impacting the LD performance after quantization, namely intra-head sensitivity and inter-head sensitivity, where a small quantization error in specific semantics can cause significant lane distortion. Thus, we propose a Selective Focus framework deployed with Semantic Guided Focus and Sensitivity Aware Selection modules, to incorporate post-processing information into PTQ reconstruction. Based on the observed intra-head sensitivity, Semantic Guided Focus is introduced to prioritize foreground-related semantics using a practical proxy. For inter-head sensitivity, we present Sensitivity Aware Selection, efficiently recognizing influential prediction heads and refining the optimization objectives at runtime. Extensive experiments have been done on a wide variety of models including keypoint-, anchor-, curve-, and segmentation-based ones. Our method produces quantized models in minutes on a single GPU and can achieve 6.4% F1 Score improvement on the CULane dataset.
Deep neural networks (DNNs) have been widely used in many artificial intelligence (AI) tasks. However, deploying them brings significant challenges due to the huge cost of memory, energy, and computation. To address these challenges, researchers have developed various model compression techniques such as model quantization and model pruning. Recently, there has been a surge in research of compression methods to achieve model efficiency while retaining the performance. Furthermore, more and more works focus on customizing the DNN hardware accelerators to better leverage the model compression techniques. In addition to efficiency, preserving security and privacy is critical for deploying DNNs. However, the vast and diverse body of related works can be overwhelming. This inspires us to conduct a comprehensive survey on recent research toward the goal of high-performance, cost-efficient, and safe deployment of DNNs. Our survey first covers the mainstream model compression techniques such as model quantization, model pruning, knowledge distillation, and optimizations of non-linear operations. We then introduce recent advances in designing hardware accelerators that can adapt to efficient model compression approaches. Additionally, we discuss how homomorphic encryption can be integrated to secure DNN deployment. Finally, we discuss several issues, such as hardware evaluation, generalization, and integration of various compression approaches. Overall, we aim to provide a big picture of efficient DNNs, from algorithm to hardware accelerators and security perspectives.
Recent advancements in large language models (LLMs) are propelling us toward artificial general intelligence with their remarkable emergent abilities and reasoning capabilities. However, the substantial computational and memory requirements limit the widespread adoption. Quantization, a key compression technique, can effectively mitigate these demands by compressing and accelerating LLMs, albeit with potential risks to accuracy. Numerous studies have aimed to minimize the accuracy loss associated with quantization. However, their quantization configurations vary from each other and cannot be fairly compared. In this paper, we present LLMC, a plug-and-play compression toolkit, to fairly and systematically explore the impact of quantization. LLMC integrates dozens of algorithms, models, and hardwares, offering high extensibility from integer to floating-point quantization, from LLM to vision-language (VLM) model, from fixed-bit to mixed precision, and from quantization to sparsification. Powered by this versatile toolkit, our benchmark covers three key aspects: calibration data, algorithms (three strategies), and data formats, providing novel insights and detailed analyses for further research and practical guidance for users. Our toolkit is available at https://github.com/ModelTC/llmc.
Due to the rising privacy concerns on sensitive client data and trained models like Transformers, secure multi-party computation (MPC) techniques are employed to enable secure inference despite attendant overhead. Existing works attempt to reduce the overhead using more MPC-friendly non-linear function approximations. However, the integration of quantization widely used in plaintext inference into the MPC domain remains unclear. To bridge this gap, we propose the framework named Ditto to enable more efficient quantization-aware secure Transformer inference. Concretely, we first incorporate an MPC-friendly quantization into Transformer inference and employ a quantization-aware distillation procedure to maintain the model utility. Then, we propose novel MPC primitives to support the type conversions that are essential in quantization and implement the quantization-aware MPC execution of secure quantized inference. This approach significantly decreases both computation and communication overhead, leading to improvements in overall efficiency. We conduct extensive experiments on Bert and GPT2 models to evaluate the performance of Ditto. The results demonstrate that Ditto is about $3.14\sim 4.40\times$ faster than MPCFormer (ICLR 2023) and $1.44\sim 2.35\times$ faster than the state-of-the-art work PUMA with negligible utility degradation.
Quantization-aware training comes with a fundamental challenge: the derivative of quantization functions such as rounding are zero almost everywhere and nonexistent elsewhere. Various differentiable approximations of quantization functions have been proposed to address this issue. In this paper, we prove that when the learning rate is sufficiently small, a large class of weight gradient estimators is equivalent with the straight through estimator (STE). Specifically, after swapping in the STE and adjusting both the weight initialization and the learning rate in SGD, the model will train in almost exactly the same way as it did with the original gradient estimator. Moreover, we show that for adaptive learning rate algorithms like Adam, the same result can be seen without any modifications to the weight initialization and learning rate. We experimentally show that these results hold for both a small convolutional model trained on the MNIST dataset and for a ResNet50 model trained on ImageNet.
Efficient deployment of Large Language Models (LLMs) requires batching multiple requests together to improve throughput. As the batch size, context length, or model size increases, the size of the key and value (KV) cache can quickly become the main contributor to GPU memory usage and the bottleneck of inference latency. Quantization has emerged as an effective technique for KV cache compression, but existing methods still fail at very low bit widths. We observe that distinct channels of a key/value activation embedding are highly inter-dependent, and the joint entropy of multiple channels grows at a slower rate than the sum of their marginal entropies. Based on this insight, we propose Coupled Quantization (CQ), which couples multiple key/value channels together to exploit their inter-dependency and encode the activations in a more information-efficient manner. Extensive experiments reveal that CQ outperforms or is competitive with existing baselines in preserving model quality. Furthermore, we demonstrate that CQ can preserve model quality with KV cache quantized down to 1-bit.
This paper introduces DeltaKWS, to the best of our knowledge, the first $\Delta$RNN-enabled fine-grained temporal sparsity-aware KWS IC for voice-controlled devices. The 65 nm prototype chip features a number of techniques to enhance performance, area, and power efficiencies, specifically: 1) a bio-inspired delta-gated recurrent neural network ($\Delta$RNN) classifier leveraging temporal similarities between neighboring feature vectors extracted from input frames and network hidden states, eliminating unnecessary operations and memory accesses; 2) an IIR BPF-based FEx that leverages mixed-precision quantization, low-cost computing structure and channel selection; 3) a 24 kB 0.6 V near-$V_\text{TH}$ weight SRAM that achieves 6.6X lower read power than the foundry-provided SRAM. From chip measurement results, we show that the DeltaKWS achieves an 11/12-class GSCD accuracy of 90.5%/89.5% respectively and energy consumption of 36 nJ/decision in 65 nm CMOS process. At 87% temporal sparsity, computing latency and energy/inference are reduced by 2.4X/3.4X, respectively. The IIR BPF-based FEx, $\Delta$RNN accelerator, and 24 kB near-$V_\text{TH}$ SRAM blocks occupy 0.084 mm$^{2}$, 0.319 mm$^{2}$, and 0.381 mm$^{2}$ respectively (0.78 mm$^{2}$ in total).
Motivated by the huge success of Transformers in the field of natural language processing (NLP), Vision Transformers (ViTs) have been rapidly developed and achieved remarkable performance in various computer vision tasks. However, their huge model sizes and intensive computations hinder ViTs' deployment on embedded devices, calling for effective model compression methods, such as quantization. Unfortunately, due to the existence of hardware-unfriendly and quantization-sensitive non-linear operations, particularly {Softmax}, it is non-trivial to completely quantize all operations in ViTs, yielding either significant accuracy drops or non-negligible hardware costs. In response to challenges associated with \textit{standard ViTs}, we focus our attention towards the quantization and acceleration for \textit{efficient ViTs}, which not only eliminate the troublesome Softmax but also integrate linear attention with low computational complexity, and propose Trio-ViT accordingly. Specifically, at the algorithm level, we develop a {tailored post-training quantization engine} taking the unique activation distributions of Softmax-free efficient ViTs into full consideration, aiming to boost quantization accuracy. Furthermore, at the hardware level, we build an accelerator dedicated to the specific Convolution-Transformer hybrid architecture of efficient ViTs, thereby enhancing hardware efficiency. Extensive experimental results consistently prove the effectiveness of our Trio-ViT framework. {Particularly, we can gain up to $\uparrow$$\mathbf{3.6}\times$, $\uparrow$$\mathbf{5.0}\times$, and $\uparrow$$\mathbf{7.3}\times$ FPS under comparable accuracy over state-of-the-art ViT accelerators, as well as $\uparrow$$\mathbf{6.0}\times$, $\uparrow$$\mathbf{1.5}\times$, and $\uparrow$$\mathbf{2.1}\times$ DSP efficiency.} Codes are available at \url{https://github.com/shihuihong214/Trio-ViT}.
Quantization can accelerate large language model (LLM) inference. Going beyond INT8 quantization, the research community is actively exploring even lower precision, such as INT4. Nonetheless, state-of-the-art INT4 quantization techniques only accelerate low-batch, edge LLM inference, failing to deliver performance gains in large-batch, cloud-based LLM serving. We uncover a critical issue: existing INT4 quantization methods suffer from significant runtime overhead (20-90%) when dequantizing either weights or partial sums on GPUs. To address this challenge, we introduce QoQ, a W4A8KV4 quantization algorithm with 4-bit weight, 8-bit activation, and 4-bit KV cache. QoQ stands for quattuor-octo-quattuor, which represents 4-8-4 in Latin. QoQ is implemented by the QServe inference library that achieves measured speedup. The key insight driving QServe is that the efficiency of LLM serving on GPUs is critically influenced by operations on low-throughput CUDA cores. Building upon this insight, in QoQ algorithm, we introduce progressive quantization that can allow low dequantization overhead in W4A8 GEMM. Additionally, we develop SmoothAttention to effectively mitigate the accuracy degradation incurred by 4-bit KV quantization. In the QServe system, we perform compute-aware weight reordering and take advantage of register-level parallelism to reduce dequantization latency. We also make fused attention memory-bound, harnessing the performance gain brought by KV4 quantization. As a result, QServe improves the maximum achievable serving throughput of Llama-3-8B by 1.2x on A100, 1.4x on L40S; and Qwen1.5-72B by 2.4x on A100, 3.5x on L40S, compared to TensorRT-LLM. Remarkably, QServe on L40S GPU can achieve even higher throughput than TensorRT-LLM on A100. Thus, QServe effectively reduces the dollar cost of LLM serving by 3x. Code is available at https://github.com/mit-han-lab/omniserve.
Channel estimation in quantized systems is challenging, particularly in low-resolution systems. In this work, we propose to leverage a Gaussian mixture model (GMM) as generative prior, capturing the channel distribution of the propagation environment, to enhance a classical estimation technique based on the expectation-maximization (EM) algorithm for one-bit quantization. Thereby, a maximum a posteriori (MAP) estimate of the most responsible mixture component is inferred for a quantized received signal, which is subsequently utilized in the EM algorithm as side information. Numerical results demonstrate the significant performance improvement of our proposed approach over both a simplistic Gaussian prior and current state-of-the-art channel estimators. Furthermore, the proposed estimation framework exhibits adaptability to higher resolution systems and alternative generative priors.