This paper proposes two nonlinear dynamics to solve constrained distributed optimization problem for resource allocation over a multi-agent network. In this setup, coupling constraint refers to resource-demand balance which is preserved at all-times. The proposed solutions can address various model nonlinearities, for example, due to quantization and/or saturation. Further, it allows to reach faster convergence or to robustify the solution against impulsive noise or uncertainties. We prove convergence over weakly connected networks using convex analysis and Lyapunov theory. Our findings show that convergence can be reached for general sign-preserving odd nonlinearity. We further propose delay-tolerant mechanisms to handle general bounded heterogeneous time-varying delays over the communication network of agents while preserving all-time feasibility. This work finds application in CPU scheduling and coverage control among others. This paper advances the state-of-the-art by addressing (i) possible nonlinearity on the agents/links, meanwhile handling (ii) resource-demand feasibility at all times, (iii) uniform-connectivity instead of all-time connectivity, and (iv) possible heterogeneous and time-varying delays. To our best knowledge, no existing work addresses contributions (i)-(iv) altogether. Simulations and comparative analysis are provided to corroborate our contributions.
With recent advancements, the wireless local area network (WLAN) or wireless fidelity (Wi-Fi) technology has been successfully utilized to realize sensing functionalities such as detection, localization, and recognition. However, the WLANs standards are developed mainly for the purpose of communication, and thus may not be able to meet the stringent requirements for emerging sensing applications. To resolve this issue, a new Task Group (TG), namely IEEE 802.11bf, has been established by the IEEE 802.11 working group, with the objective of creating a new amendment to the WLAN standard to meet advanced sensing requirements while minimizing the effect on communications. This paper provides a comprehensive overview on the up-to-date efforts in the IEEE 802.11bf TG. First, we introduce the definition of the 802.11bf amendment and its formation and standardization timeline. Next, we discuss the WLAN sensing use cases with the corresponding key performance indicator (KPI) requirements. After reviewing previous WLAN sensing research based on communication-oriented WLAN standards, we identify their limitations and underscore the practical need for the new sensing-oriented amendment in 802.11bf. Furthermore, we discuss the WLAN sensing framework and procedure used for measurement acquisition, by considering both sensing at sub-7GHz and directional multi-gigabit (DMG) sensing at 60 GHz, respectively, and address their shared features, similarities, and differences. In addition, we present various candidate technical features for IEEE 802.11bf, including waveform/sequence design, feedback types, as well as quantization and compression techniques. We also describe the methodologies and the channel modeling used by the IEEE 802.11bf TG for evaluation. Finally, we discuss the challenges and future research directions to motivate more research endeavors towards this field in details.
Assessing the condition and visibility of veins is a crucial step before obtaining intravenous access in the antecubital fossa, which is a common procedure to draw blood or administer intravenous therapies (IV therapies). Even though medical practitioners are highly skilled at intravenous cannulation, they usually struggle to perform the procedure in patients with low visible veins due to fluid retention, age, overweight, dark skin tone, or diabetes. Recently, several investigations proposed combining Near Infrared (NIR) imaging and deep learning (DL) techniques for forearm vein segmentation. Although they have demonstrated compelling results, their use has been rather limited owing to the portability and precision requirements to perform venipuncture. In this paper, we aim to contribute to bridging this gap using three strategies. First, we introduce a new NIR-based forearm vein segmentation dataset of 2,016 labelled images collected from 1,008 subjects with low visible veins. Second, we propose a modified U-Net architecture that locates veins specifically in the antecubital fossa region of the examined patient. Finally, a compressed version of the proposed architecture was deployed inside a bespoke, portable vein finder device after testing four common embedded microcomputers and four common quantization modalities. Experimental results showed that the model compressed with Dynamic Range Quantization and deployed on a Raspberry Pi 4B card produced the best execution time and precision balance, with 5.14 FPS and 0.957 of latency and Intersection over Union (IoU), respectively. These results show promising performance inside a resource-restricted low-cost device.
Quantization techniques are pivotal in reducing the memory and computational demands of deep neural network inference. Existing solutions, such as ZeroQuant, offer dynamic quantization for models like BERT and GPT but overlook crucial memory-bounded operators and the complexities of per-token quantization. Addressing these gaps, we present a novel, fully hardware-enhanced robust optimized post-training W8A8 quantization framework, ZeroQuant-HERO. This framework uniquely integrates both memory bandwidth and compute-intensive operators, aiming for optimal hardware performance. Additionally, it offers flexibility by allowing specific INT8 modules to switch to FP16/BF16 mode, enhancing accuracy.
Understanding neural networks is challenging in part because of the dense, continuous nature of their hidden states. We explore whether we can train neural networks to have hidden states that are sparse, discrete, and more interpretable by quantizing their continuous features into what we call codebook features. Codebook features are produced by finetuning neural networks with vector quantization bottlenecks at each layer, producing a network whose hidden features are the sum of a small number of discrete vector codes chosen from a larger codebook. Surprisingly, we find that neural networks can operate under this extreme bottleneck with only modest degradation in performance. This sparse, discrete bottleneck also provides an intuitive way of controlling neural network behavior: first, find codes that activate when the desired behavior is present, then activate those same codes during generation to elicit that behavior. We validate our approach by training codebook Transformers on several different datasets. First, we explore a finite state machine dataset with far more hidden states than neurons. In this setting, our approach overcomes the superposition problem by assigning states to distinct codes, and we find that we can make the neural network behave as if it is in a different state by activating the code for that state. Second, we train Transformer language models with up to 410M parameters on two natural language datasets. We identify codes in these models representing diverse, disentangled concepts (ranging from negative emotions to months of the year) and find that we can guide the model to generate different topics by activating the appropriate codes during inference. Overall, codebook features appear to be a promising unit of analysis and control for neural networks and interpretability. Our codebase and models are open-sourced at https://github.com/taufeeque9/codebook-features.
Regression tasks in computer vision, such as age estimation or counting, are often formulated into classification by quantizing the target space into classes. Yet real-world data is often imbalanced -- the majority of training samples lie in a head range of target values, while a minority of samples span a usually larger tail range. By selecting the class quantization, one can adjust imbalanced regression targets into balanced classification outputs, though there are trade-offs in balancing classification accuracy and quantization error. To improve regression performance over the entire range of data, we propose to construct hierarchical classifiers for solving imbalanced regression tasks. The fine-grained classifiers limit the quantization error while being modulated by the coarse predictions to ensure high accuracy. Standard hierarchical classification approaches, however, when applied to the regression problem, fail to ensure that predicted ranges remain consistent across the hierarchy. As such, we propose a range-preserving distillation process that can effectively learn a single classifier from the set of hierarchical classifiers. Our novel hierarchical classification adjustment (HCA) for imbalanced regression shows superior results on three diverse tasks: age estimation, crowd counting and depth estimation. We will release the source code upon acceptance.
We propose a novel multi-bit box-free watermarking method for the protection of Intellectual Property Rights (IPR) of GANs with improved robustness against white-box attacks like fine-tuning, pruning, quantization, and surrogate model attacks. The watermark is embedded by adding an extra watermarking loss term during GAN training, ensuring that the images generated by the GAN contain an invisible watermark that can be retrieved by a pre-trained watermark decoder. In order to improve the robustness against white-box model-level attacks, we make sure that the model converges to a wide flat minimum of the watermarking loss term, in such a way that any modification of the model parameters does not erase the watermark. To do so, we add random noise vectors to the parameters of the generator and require that the watermarking loss term is as invariant as possible with respect to the presence of noise. This procedure forces the generator to converge to a wide flat minimum of the watermarking loss. The proposed method is architectureand dataset-agnostic, thus being applicable to many different generation tasks and models, as well as to CNN-based image processing architectures. We present the results of extensive experiments showing that the presence of the watermark has a negligible impact on the quality of the generated images, and proving the superior robustness of the watermark against model modification and surrogate model attacks.
The pre-training architectures of large language models encompass various types, including autoencoding models, autoregressive models, and encoder-decoder models. We posit that any modality can potentially benefit from a large language model, as long as it undergoes vector quantization to become discrete tokens. Inspired by GLM, we propose a General Point Model (GPM) which seamlessly integrates autoencoding and autoregressive tasks in point cloud transformer. This model is versatile, allowing fine-tuning for downstream point cloud representation tasks, as well as unconditional and conditional generation tasks. GPM enhances masked prediction in autoencoding through various forms of mask padding tasks, leading to improved performance in point cloud understanding. Additionally, GPM demonstrates highly competitive results in unconditional point cloud generation tasks, even exhibiting the potential for conditional generation tasks by modifying the input's conditional information. Compared to models like Point-BERT, MaskPoint and PointMAE, our GPM achieves superior performance in point cloud understanding tasks. Furthermore, the integration of autoregressive and autoencoding within the same transformer underscores its versatility across different downstream tasks.
Transmitting data using the phases on reconfigurable intelligent surfaces (RIS) is a promising solution for future energy-efficient communication systems. Recent work showed that a virtual phased massive multiuser multiple-input-multiple-out (MIMO) transmitter can be formed using only one active antenna and a large passive RIS. In this paper, we are interested in using such a system to perform MIMO downlink precoding. In this context, we may not be able to apply conventional MIMO precoding schemes, such as the simple zero-forcing (ZF) scheme, and we typically need to design the phase signals by solving optimization problems with constant modulus constraints or with discrete phase constraints, which pose challenges with high computational complexities. In this work, we propose an alternative approach based on Sigma-Delta ($\Sigma\Delta$) modulation, which is classically famous for its noise-shaping ability. Specifically, first-order $\Sigma\Delta$ modulation is applied in the spatial domain to handle phase quantization in generating constant envelope signals. Under some mild assumptions, the proposed phased $\Sigma\Delta$ modulator allows us to use the ZF scheme to synthesize the RIS reflection phases with negligible complexity. The proposed approach is empirically shown to achieve comparable bit error rate performance to the unquantized ZF scheme.
Low-precision training has emerged as a promising low-cost technique to enhance the training efficiency of deep neural networks without sacrificing much accuracy. Its Bayesian counterpart can further provide uncertainty quantification and improved generalization accuracy. This paper investigates low-precision sampling via Stochastic Gradient Hamiltonian Monte Carlo (SGHMC) with low-precision and full-precision gradient accumulators for both strongly log-concave and non-log-concave distributions. Theoretically, our results show that, to achieve $\epsilon$-error in the 2-Wasserstein distance for non-log-concave distributions, low-precision SGHMC achieves quadratic improvement ($\widetilde{\mathbf{O}}\left({\epsilon^{-2}{\mu^*}^{-2}\log^2\left({\epsilon^{-1}}\right)}\right)$) compared to the state-of-the-art low-precision sampler, Stochastic Gradient Langevin Dynamics (SGLD) ($\widetilde{\mathbf{O}}\left({{\epsilon}^{-4}{\lambda^{*}}^{-1}\log^5\left({\epsilon^{-1}}\right)}\right)$). Moreover, we prove that low-precision SGHMC is more robust to the quantization error compared to low-precision SGLD due to the robustness of the momentum-based update w.r.t. gradient noise. Empirically, we conduct experiments on synthetic data, and {MNIST, CIFAR-10 \& CIFAR-100} datasets, which validate our theoretical findings. Our study highlights the potential of low-precision SGHMC as an efficient and accurate sampling method for large-scale and resource-limited machine learning.
In the realm of music AI, arranging rich and structured multi-track accompaniments from a simple lead sheet presents significant challenges. Such challenges include maintaining track cohesion, ensuring long-term coherence, and optimizing computational efficiency. In this paper, we introduce a novel system that leverages prior modelling over disentangled style factors to address these challenges. Our method presents a two-stage process: initially, a piano arrangement is derived from the lead sheet by retrieving piano texture styles; subsequently, a multi-track orchestration is generated by infusing orchestral function styles into the piano arrangement. Our key design is the use of vector quantization and a unique multi-stream Transformer to model the long-term flow of the orchestration style, which enables flexible, controllable, and structured music generation. Experiments show that by factorizing the arrangement task into interpretable sub-stages, our approach enhances generative capacity while improving efficiency. Additionally, our system supports a variety of music genres and provides style control at different composition hierarchies. We further show that our system achieves superior coherence, structure, and overall arrangement quality compared to existing baselines.
We propose LLM-FP4 for quantizing both weights and activations in large language models (LLMs) down to 4-bit floating-point values, in a post-training manner. Existing post-training quantization (PTQ) solutions are primarily integer-based and struggle with bit widths below 8 bits. Compared to integer quantization, floating-point (FP) quantization is more flexible and can better handle long-tail or bell-shaped distributions, and it has emerged as a default choice in many hardware platforms. One characteristic of FP quantization is that its performance largely depends on the choice of exponent bits and clipping range. In this regard, we construct a strong FP-PTQ baseline by searching for the optimal quantization parameters. Furthermore, we observe a high inter-channel variance and low intra-channel variance pattern in activation distributions, which adds activation quantization difficulty. We recognize this pattern to be consistent across a spectrum of transformer models designed for diverse tasks, such as LLMs, BERT, and Vision Transformer models. To tackle this, we propose per-channel activation quantization and show that these additional scaling factors can be reparameterized as exponential biases of weights, incurring a negligible cost. Our method, for the first time, can quantize both weights and activations in the LLaMA-13B to only 4-bit and achieves an average score of 63.1 on the common sense zero-shot reasoning tasks, which is only 5.8 lower than the full-precision model, significantly outperforming the previous state-of-the-art by 12.7 points. Code is available at: https://github.com/nbasyl/LLM-FP4.
Recently, neural networks have improved MinSum message-passing decoders for low-density parity-check (LDPC) codes by multiplying or adding weights to the messages, where the weights are determined by a neural network. The neural network complexity to determine distinct weights for each edge is high, often limiting the application to relatively short LDPC codes. Furthermore, storing separate weights for every edge and every iteration can be a burden for hardware implementations. To reduce neural network complexity and storage requirements, this paper proposes a family of weight-sharing schemes that use the same weight for edges that have the same check node degree and/or variable node degree. Our simulation results show that node-degree-based weight-sharing can deliver the same performance requiring distinct weights for each node. This paper also combines these degree-specific neural weights with a reconstruction-computation-quantization (RCQ) decoder to produce a weighted RCQ (W-RCQ) decoder. The W-RCQ decoder with node-degree-based weight sharing has a reduced hardware requirement compared with the original RCQ decoder. As an additional contribution, this paper identifies and resolves a gradient explosion issue that can arise when training neural LDPC decoders.
Representation Learning on Knowledge Graphs (KGs) is essential for downstream tasks. The dominant approach, KG Embedding (KGE), represents entities with independent vectors and faces the scalability challenge. Recent studies propose an alternative way for parameter efficiency, which represents entities by composing entity-corresponding codewords matched from predefined small-scale codebooks. We refer to the process of obtaining corresponding codewords of each entity as entity quantization, for which previous works have designed complicated strategies. Surprisingly, this paper shows that simple random entity quantization can achieve similar results to current strategies. We analyze this phenomenon and reveal that entity codes, the quantization outcomes for expressing entities, have higher entropy at the code level and Jaccard distance at the codeword level under random entity quantization. Therefore, different entities become more easily distinguished, facilitating effective KG representation. The above results show that current quantization strategies are not critical for KG representation, and there is still room for improvement in entity distinguishability beyond current strategies. The code to reproduce our results is available at https://github.com/JiaangL/RandomQuantization.
This paper considers the context of multiuser massive MIMO downlink precoding with low-resolution digital-to-analog converters (DACs) at the transmitter. This subject is motivated by the consideration that it is expensive to employ high-resolution DACs for practical massive MIMO implementations. The challenge with using low-resolution DACs is to overcome the detrimental quantization error effects. Recently, spatial Sigma-Delta modulation has arisen as a viable way to put quantization errors under control. This approach takes insight from temporal Sigma-Delta modulation in classical DAC studies. Assuming a 1D uniform linear transmit antenna array, the principle is to shape the quantization errors in space such that the shaped quantization errors are pushed away from the user-serving angle sector. In the previous studies, spatial Sigma-Delta modulation was performed by direct application of the basic first- and second-order modulators from the Sigma-Delta literature. In this paper, we develop a general Sigma-Delta modulator design framework for any given order, for any given number of quantization levels, and for any given angle sector. We formulate our design as a problem of maximizing the signal-to-quantization-and-noise ratios experienced by the users. The formulated problem is convex and can be efficiently solved by available solvers. Our proposed framework offers the alternative option of focused quantization error suppression in accordance with channel state information. Our framework can also be extended to 2D planar transmit antenna arrays. We perform numerical study under different operating conditions, and the numerical results suggest that, given a moderate number of quantization levels, say, 5 to 7 levels, our optimization-based Sigma-Delta modulation schemes can lead to bit error rate performance close to that of the unquantized counterpart.
Federated Learning (FL) is a promising distributed method for edge-level machine learning, particularly for privacysensitive applications such as those in military and medical domains, where client data cannot be shared or transferred to a cloud computing server. In many use-cases, communication cost is a major challenge in FL due to its natural intensive network usage. Client devices, such as smartphones or Internet of Things (IoT) nodes, have limited resources in terms of energy, computation, and memory. To address these hardware constraints, lightweight models and compression techniques such as pruning and quantization are commonly adopted in centralised paradigms. In this paper, we investigate the impact of compression techniques on FL for a typical image classification task. Going further, we demonstrate that a straightforward method can compresses messages up to 50% while having less than 1% of accuracy loss, competing with state-of-the-art techniques.
Recent advancements in implicit neural representations have contributed to high-fidelity surface reconstruction and photorealistic novel view synthesis. However, the computational complexity inherent in these methodologies presents a substantial impediment, constraining the attainable frame rates and resolutions in practical applications. In response to this predicament, we propose VQ-NeRF, an effective and efficient pipeline for enhancing implicit neural representations via vector quantization. The essence of our method involves reducing the sampling space of NeRF to a lower resolution and subsequently reinstating it to the original size utilizing a pre-trained VAE decoder, thereby effectively mitigating the sampling time bottleneck encountered during rendering. Although the codebook furnishes representative features, reconstructing fine texture details of the scene remains challenging due to high compression rates. To overcome this constraint, we design an innovative multi-scale NeRF sampling scheme that concurrently optimizes the NeRF model at both compressed and original scales to enhance the network's ability to preserve fine details. Furthermore, we incorporate a semantic loss function to improve the geometric fidelity and semantic coherence of our 3D reconstructions. Extensive experiments demonstrate the effectiveness of our model in achieving the optimal trade-off between rendering quality and efficiency. Evaluation on the DTU, BlendMVS, and H3DS datasets confirms the superior performance of our approach.
A deep autoencoder (DAE)-based structure for endto-end communication over the two-user Z-interference channel (ZIC) with finite-alphabet inputs is designed in this paper. The proposed structure jointly optimizes the two encoder/decoder pairs and generates interference-aware constellations that dynamically adapt their shape based on interference intensity to minimize the bit error rate (BER). An in-phase/quadrature-phase (I/Q) power allocation layer is introduced in the DAE to guarantee an average power constraint and enable the architecture to generate constellations with nonuniform shapes. This brings further gain compared to standard uniform constellations such as quadrature amplitude modulation. The proposed structure is then extended to work with imperfect channel state information (CSI). The CSI imperfection due to both the estimation and quantization errors are examined. The performance of the DAEZIC is compared with two baseline methods, i.e., standard and rotated constellations. The proposed structure significantly enhances the performance of the ZIC both for the perfect and imperfect CSI. Simulation results show that the improvement is achieved in all interference regimes (weak, moderate, and strong) and consistently increases with the signal-to-noise ratio (SNR). For example, more than an order of magnitude BER reduction is obtained with respect to the most competitive conventional method at weak interference when SNR>15dB and two bits per symbol are transmitted. The improvements reach about two orders of magnitude when quantization error exists, indicating that the DAE-ZIC is more robust to the interference compared to the conventional methods.
Quantization is a promising approach for reducing memory overhead and accelerating inference, especially in large pre-trained language model (PLM) scenarios. While having no access to original training data due to security and privacy concerns has emerged the demand for zero-shot quantization. Most of the cutting-edge zero-shot quantization methods primarily 1) apply to computer vision tasks, and 2) neglect of overfitting problem in the generative adversarial learning process, leading to sub-optimal performance. Motivated by this, we propose a novel zero-shot sharpness-aware quantization (ZSAQ) framework for the zero-shot quantization of various PLMs. The key algorithm in solving ZSAQ is the SAM-SGA optimization, which aims to improve the quantization accuracy and model generalization via optimizing a minimax problem. We theoretically prove the convergence rate for the minimax optimization problem and this result can be applied to other nonconvex-PL minimax optimization frameworks. Extensive experiments on 11 tasks demonstrate that our method brings consistent and significant performance gains on both discriminative and generative PLMs, i.e., up to +6.98 average score. Furthermore, we empirically validate that our method can effectively improve the model generalization.
Physical layer key generation based on reciprocal and random wireless channels has been an attractive solution for securing resource-constrained low-power wide-area networks (LPWANs). When quantizing channel measurements, namely received signal strength indicator (RSSI), into key bits, the existing works mainly adopt fixed quantization levels and guard band parameters, which fail to fully extract keys from RSSI measurements. In this paper, we propose a novel adaptive quantization scheme for key generation in LPWANs, taking LoRa as a case study. The proposed adaptive quantization scheme can dynamically adjust the quantization parameters according to the randomness of RSSI measurements estimated by Lempel-Ziv complexity (LZ76), while ensuring a predefined key disagreement ratio (KDR). Specifically, our scheme uses pre-trained linear regression models to determine the appropriate quantization level and guard band parameter for each segment of RSSI measurements. Moreover, we propose a guard band parameter calibration scheme during information reconciliation during real-time key generation operation. Experimental evaluations using LoRa devices show that the proposed adaptive quantization scheme outperforms the benchmark differential quantization and fixed quantization with up to 2.35$\times$ and 1.51$\times$ key generation rate (KGR) gains, respectively.