arXiv subjects

Yifu Ding

Publications and source records attributed to Yifu Ding.

At least 37 records · Page 2Linked to original sources

arXiv2024

DB-LLM: Accurate Dual-Binarization for Efficient LLMs

Large language models (LLMs) have significantly advanced the field of natural language processing, while the expensive memory and computation consumption impede their practical deployment. Quantization emerges as one of the most effective methods for improving the computational efficiency of LLMs. However, existing ultra-low-bit quantization always causes severe accuracy drops. In this paper, we empirically relieve the micro and macro characteristics of ultra-low bit quantization and present a novel Dual-Binarization method for LLMs, namely DB-LLM. For the micro-level, we take both the accuracy advantage of 2-bit-width and the efficiency advantage of binarization into account, introducing Flexible Dual Binarization (FDB). By splitting 2-bit quantized weights into two independent sets of binaries, FDB ensures the accuracy of representations and introduces flexibility, utilizing the efficient bitwise operations of binarization while retaining the inherent high sparsity of ultra-low bit quantization. For the macro-level, we find the distortion that exists in the prediction of LLM after quantization, which is specified as the deviations related to the ambiguity of samples. We propose the Deviation-Aware Distillation (DAD) method, enabling the model to focus differently on various samples. Comprehensive experiments show that our DB-LLM not only significantly surpasses the current State-of-The-Art (SoTA) in ultra-low bit quantization (eg, perplexity decreased from 9.64 to 7.23), but also achieves an additional 20\% reduction in computational consumption compared to the SOTA method under the same bit-width. Our code will be released soon.

cs.LG↗

arXiv2024

Repurposing Coal Power Plants into Thermal Energy Storage for Supporting Zero-carbon Data Centers

Coal power plants will need to be phased out and face stranded asset risks under the net-zero energy system transition. Repurposing coal power plants could recoup profits and reduce carbon emissions using the existing infrastructure and grid connections. This paper investigates a retrofitting strategy that turns coal power plants into thermal energy storage (TES) and zero-carbon data centers (DCs). The proposed capacity expansion model considers the co-locations of DCs, local renewablewith the system-generation, andlevel coal retir energy storage ement and retrofitting. We optimize the DC system configurations under the hourly-matching carbon policy and flexible operations. Results show that under hourly-matching carbon constraints, the retrofitted TES could complement the operations of lithium-ion batteries (LIBs) to reduce system costs. This could render DCs with optimal co-located renewable generations and energy storage more cost-effective than unconstrained DCs.

eess.SY↗

arXiv2023

Hippocampus Substructure Segmentation Using Morphological Vision Transformer Learning

Background: The hippocampus plays a crucial role in memory and cognition. Because of the associated toxicity from whole brain radiotherapy, more advanced treatment planning techniques prioritize hippocampal avoidance, which depends on an accurate segmentation of the small and complexly shaped hippocampus. Purpose: To achieve accurate segmentation of the anterior and posterior regions of the hippocampus from T1 weighted (T1w) MRI images, we developed a novel model, Hippo-Net, which uses a mutually enhanced strategy. Methods: The proposed model consists of two major parts: 1) a localization model is used to detect the volume-of-interest (VOI) of hippocampus. 2) An end-to-end morphological vision transformer network is used to perform substructures segmentation within the hippocampus VOI. A total of 260 T1w MRI datasets were used in this study. We conducted a five-fold cross-validation on the first 200 T1w MR images and then performed a hold-out test on the remaining 60 T1w MR images with the model trained on the first 200 images. Results: In five-fold cross-validation, the DSCs were 0.900+-0.029 and 0.886+-0.031for the hippocampus proper and parts of the subiculum, respectively. The MSD were 0.426+-0.115mm and 0.401+-0.100 mm for the hippocampus proper and parts of the subiculum, respectively. Conclusions: The proposed method showed great promise in automatically delineating hippocampus substructures on T1w MRI images. It may facilitate the current clinical workflow and reduce the physician effort.

physics.med-ph↗

arXiv2023

BiBench: Benchmarking and Analyzing Network Binarization

Network binarization emerges as one of the most promising compression approaches offering extraordinary computation and memory savings by minimizing the bit-width. However, recent research has shown that applying existing binarization algorithms to diverse tasks, architectures, and hardware in realistic scenarios is still not straightforward. Common challenges of binarization, such as accuracy degradation and efficiency limitation, suggest that its attributes are not fully understood. To close this gap, we present BiBench, a rigorously designed benchmark with in-depth analysis for network binarization. We first carefully scrutinize the requirements of binarization in the actual production and define evaluation tracks and metrics for a comprehensive and fair investigation. Then, we evaluate and analyze a series of milestone binarization algorithms that function at the operator level and with extensive influence. Our benchmark reveals that 1) the binarized operator has a crucial impact on the performance and deployability of binarized networks; 2) the accuracy of binarization varies significantly across different learning tasks and neural architectures; 3) binarization has demonstrated promising efficiency potential on edge devices despite the limited hardware support. The results and analysis also lead to a promising paradigm for accurate and efficient binarization. We believe that BiBench will contribute to the broader adoption of binarization and serve as a foundation for future research. The code for our BiBench is released https://github.com/htqin/BiBench .

cs.CV↗

arXiv2023

Deep Learning in MRI-guided Radiation Therapy: A Systematic Review

MRI-guided radiation therapy (MRgRT) offers a precise and adaptive approach to treatment planning. Deep learning applications which augment the capabilities of MRgRT are systematically reviewed. MRI-guided radiation therapy offers a precise, adaptive approach to treatment planning. Deep learning applications which augment the capabilities of MRgRT are systematically reviewed with emphasis placed on underlying methods. Studies are further categorized into the areas of segmentation, synthesis, radiomics, and real time MRI. Finally, clinical implications, current challenges, and future directions are discussed.

physics.med-ph↗

arXiv2023

Towards Accurate Post-Training Quantization for Vision Transformer

Vision transformer emerges as a potential architecture for vision tasks. However, the intense computation and non-negligible delay hinder its application in the real world. As a widespread model compression technique, existing post-training quantization methods still cause severe performance drops. We find the main reasons lie in (1) the existing calibration metric is inaccurate in measuring the quantization influence for extremely low-bit representation, and (2) the existing quantization paradigm is unfriendly to the power-law distribution of Softmax. Based on these observations, we propose a novel Accurate Post-training Quantization framework for Vision Transformer, namely APQ-ViT. We first present a unified Bottom-elimination Blockwise Calibration scheme to optimize the calibration metric to perceive the overall quantization disturbance in a blockwise manner and prioritize the crucial quantization errors that influence more on the final output. Then, we design a Matthew-effect Preserving Quantization for Softmax to maintain the power-law character and keep the function of the attention mechanism. Comprehensive experiments on large-scale classification and detection datasets demonstrate that our APQ-ViT surpasses the existing post-training quantization methods by convincing margins, especially in lower bit-width settings (e.g., averagely up to 5.17% improvement for classification and 24.43% for detection on W4A4). We also highlight that APQ-ViT enjoys versatility and works well on diverse transformer variants.

cs.CV↗

arXiv2023

BiFSMNv2: Pushing Binary Neural Networks for Keyword Spotting to Real-Network Performance

Deep neural networks, such as the Deep-FSMN, have been widely studied for keyword spotting (KWS) applications while suffering expensive computation and storage. Therefore, network compression technologies like binarization are studied to deploy KWS models on edge. In this paper, we present a strong yet efficient binary neural network for KWS, namely BiFSMNv2, pushing it to the real-network accuracy performance. First, we present a Dual-scale Thinnable 1-bit-Architecture to recover the representation capability of the binarized computation units by dual-scale activation binarization and liberate the speedup potential from an overall architecture perspective. Second, we also construct a Frequency Independent Distillation scheme for KWS binarization-aware training, which distills the high and low-frequency components independently to mitigate the information mismatch between full-precision and binarized representations. Moreover, we propose the Learning Propagation Binarizer, a general and efficient binarizer that enables the forward and backward propagation of binary KWS networks to be continuously improved through learning. We implement and deploy the BiFSMNv2 on ARMv8 real-world hardware with a novel Fast Bitwise Computation Kernel, which is proposed to fully utilize registers and increase instruction throughput. Comprehensive experiments show our BiFSMNv2 outperforms existing binary networks for KWS by convincing margins across different datasets and achieves comparable accuracy with the full-precision networks (only a tiny 1.51% drop on Speech Commands V1-12). We highlight that benefiting from the compact architecture and optimized hardware kernel, BiFSMNv2 can achieve an impressive 25.1x speedup and 20.2x storage-saving on edge hardware.

cs.CL↗

arXiv2022

Diverse Sample Generation: Pushing the Limit of Generative Data-free Quantization

Generative data-free quantization emerges as a practical compression approach that quantizes deep neural networks to low bit-width without accessing the real data. This approach generates data utilizing batch normalization (BN) statistics of the full-precision networks to quantize the networks. However, it always faces the serious challenges of accuracy degradation in practice. We first give a theoretical analysis that the diversity of synthetic samples is crucial for the data-free quantization, while in existing approaches, the synthetic data completely constrained by BN statistics experimentally exhibit severe homogenization at distribution and sample levels. This paper presents a generic Diverse Sample Generation (DSG) scheme for the generative data-free quantization, to mitigate detrimental homogenization. We first slack the statistics alignment for features in the BN layer to relax the distribution constraint. Then, we strengthen the loss impact of the specific BN layers for different samples and inhibit the correlation among samples in the generation process, to diversify samples from the statistical and spatial perspectives, respectively. Comprehensive experiments show that for large-scale image classification tasks, our DSG can consistently quantization performance on different neural architectures, especially under ultra-low bit-width. And data diversification caused by our DSG brings a general gain to various quantization-aware training and post-training quantization approaches, demonstrating its generality and effectiveness.

cs.CV↗

arXiv2022

BiFSMN: Binary Neural Network for Keyword Spotting

The deep neural networks, such as the Deep-FSMN, have been widely studied for keyword spotting (KWS) applications. However, computational resources for these networks are significantly constrained since they usually run on-call on edge devices. In this paper, we present BiFSMN, an accurate and extreme-efficient binary neural network for KWS. We first construct a High-frequency Enhancement Distillation scheme for the binarization-aware training, which emphasizes the high-frequency information from the full-precision network's representation that is more crucial for the optimization of the binarized network. Then, to allow the instant and adaptive accuracy-efficiency trade-offs at runtime, we also propose a Thinnable Binarization Architecture to further liberate the acceleration potential of the binarized network from the topology perspective. Moreover, we implement a Fast Bitwise Computation Kernel for BiFSMN on ARMv8 devices which fully utilizes registers and increases instruction throughput to push the limit of deployment efficiency. Extensive experiments show that BiFSMN outperforms existing binarization methods by convincing margins on various datasets and is even comparable with the full-precision counterpart (e.g., less than 3% drop on Speech Commands V1-12). We highlight that benefiting from the thinnable architecture and the optimized 1-bit implementation, BiFSMN can achieve an impressive 22.3x speedup and 15.5x storage-saving on real-world edge hardware. Our code is released at https://github.com/htqin/BiFSMN.

cs.CL↗

arXiv2022

Distribution-sensitive Information Retention for Accurate Binary Neural Network

Model binarization is an effective method of compressing neural networks and accelerating their inference process. However, a significant performance gap still exists between the 1-bit model and the 32-bit one. The empirical study shows that binarization causes a great loss of information in the forward and backward propagation. We present a novel Distribution-sensitive Information Retention Network (DIR-Net) that retains the information in the forward and backward propagation by improving internal propagation and introducing external representations. The DIR-Net mainly relies on three technical contributions: (1) Information Maximized Binarization (IMB): minimizing the information loss and the binarization error of weights/activations simultaneously by weight balance and standardization; (2) Distribution-sensitive Two-stage Estimator (DTE): retaining the information of gradients by distribution-sensitive soft approximation by jointly considering the updating capability and accurate gradient; (3) Representation-align Binarization-aware Distillation (RBD): retaining the representation information by distilling the representations between full-precision and binarized networks. The DIR-Net investigates both forward and backward processes of BNNs from the unified information perspective, thereby providing new insight into the mechanism of network binarization. The three techniques in our DIR-Net are versatile and effective and can be applied in various structures to improve BNNs. Comprehensive experiments on the image classification and objective detection tasks show that our DIR-Net consistently outperforms the state-of-the-art binarization approaches under mainstream and compact architectures, such as ResNet, VGG, EfficientNet, DARTS, and MobileNet. Additionally, we conduct our DIR-Net on real-world resource-limited devices which achieves 11.1x storage saving and 5.4x speedup.

cs.CV↗

arXiv2022

BiBERT: Accurate Fully Binarized BERT

The large pre-trained BERT has achieved remarkable performance on Natural Language Processing (NLP) tasks but is also computation and memory expensive. As one of the powerful compression approaches, binarization extremely reduces the computation and memory consumption by utilizing 1-bit parameters and bitwise operations. Unfortunately, the full binarization of BERT (i.e., 1-bit weight, embedding, and activation) usually suffer a significant performance drop, and there is rare study addressing this problem. In this paper, with the theoretical justification and empirical analysis, we identify that the severe performance drop can be mainly attributed to the information degradation and optimization direction mismatch respectively in the forward and backward propagation, and propose BiBERT, an accurate fully binarized BERT, to eliminate the performance bottlenecks. Specifically, BiBERT introduces an efficient Bi-Attention structure for maximizing representation information statistically and a Direction-Matching Distillation (DMD) scheme to optimize the full binarized BERT accurately. Extensive experiments show that BiBERT outperforms both the straightforward baseline and existing state-of-the-art quantized BERTs with ultra-low bit activations by convincing margins on the NLP benchmark. As the first fully binarized BERT, our method yields impressive 56.3 times and 31.2 times saving on FLOPs and model size, demonstrating the vast advantages and potential of the fully binarized BERT model in real-world resource-constrained scenarios.

cs.CL↗

arXiv2022

Distributionally Robust Joint Chance-Constrained Optimization for Networked Microgrids Considering Contingencies and Renewable Uncertainty

In light of a reliable and resilient power system under extreme weather and natural disasters, networked microgrids integrating local renewable resources have been adopted extensively to supply demands when the main utility experiences blackouts. However, the stochastic nature of renewables and unpredictable contingencies are difficult to address with the deterministic energy management framework. The paper proposes a comprehensive distributionally robust joint chance-constrained (DR-JCC) framework that incorporates microgrid island, power flow, distributed batteries and voltage control constraints. All chance constraints are solved jointly and each one is assigned to an optimized violation rate. To highlight, the JCC problem with the optimized violation rates has been recognized to be NP-hard and challenging to be solved. This paper proposes a novel evolutionary algorithm that successfully tackles the problem and reduces the solution conservativeness (i.e. operation cost) by around 50% comparing with the baseline Bonferroni Approximation. Considering the imperfect solar power forecast, we construct three data-driven ambiguity sets to model uncertain forecast error distributions. The solution is thus robust for any distribution in sets with the shared moment and shape assumptions. The proposed method is validated by robustness tests based on those sets and firmly secures the solution robustness.

eess.SY↗

arXiv2021

Diversifying Sample Generation for Accurate Data-Free Quantization

Quantization has emerged as one of the most prevalent approaches to compress and accelerate neural networks. Recently, data-free quantization has been widely studied as a practical and promising solution. It synthesizes data for calibrating the quantized model according to the batch normalization (BN) statistics of FP32 ones and significantly relieves the heavy dependency on real training data in traditional quantization methods. Unfortunately, we find that in practice, the synthetic data identically constrained by BN statistics suffers serious homogenization at both distribution level and sample level and further causes a significant performance drop of the quantized model. We propose Diverse Sample Generation (DSG) scheme to mitigate the adverse effects caused by homogenization. Specifically, we slack the alignment of feature statistics in the BN layer to relax the constraint at the distribution level and design a layerwise enhancement to reinforce specific layers for different data samples. Our DSG scheme is versatile and even able to be applied to the state-of-the-art post-training quantization method like AdaRound. We evaluate the DSG scheme on the large-scale image classification task and consistently obtain significant improvements over various network architectures and quantization methods, especially when quantized to lower bits (e.g., up to 22% improvement on W4A4). Moreover, benefiting from the enhanced diversity, models calibrated by synthetic data perform close to those calibrated by real data and even outperform them on W4A4.

cs.CV↗

arXiv2021

BiPointNet: Binary Neural Network for Point Clouds

To alleviate the resource constraint for real-time point cloud applications that run on edge devices, in this paper we present BiPointNet, the first model binarization approach for efficient deep learning on point clouds. We discover that the immense performance drop of binarized models for point clouds mainly stems from two challenges: aggregation-induced feature homogenization that leads to a degradation of information entropy, and scale distortion that hinders optimization and invalidates scale-sensitive structures. With theoretical justifications and in-depth analysis, our BiPointNet introduces Entropy-Maximizing Aggregation (EMA) to modulate the distribution before aggregation for the maximum information entropy, and Layer-wise Scale Recovery (LSR) to efficiently restore feature representation capacity. Extensive experiments show that BiPointNet outperforms existing binarization methods by convincing margins, at the level even comparable with the full precision counterpart. We highlight that our techniques are generic, guaranteeing significant improvements on various fundamental tasks and mainstream backbones. Moreover, BiPointNet gives an impressive 14.7x speedup and 18.9x storage saving on real-world resource-constrained devices.

cs.CV↗

arXiv2021

Multi-Pretext Attention Network for Few-shot Learning with Self-supervision

Few-shot learning is an interesting and challenging study, which enables machines to learn from few samples like humans. Existing studies rarely exploit auxiliary information from large amount of unlabeled data. Self-supervised learning is emerged as an efficient method to utilize unlabeled data. Existing self-supervised learning methods always rely on the combination of geometric transformations for the single sample by augmentation, while seriously neglect the endogenous correlation information among different samples that is the same important for the task. In this work, we propose a Graph-driven Clustering (GC), a novel augmentation-free method for self-supervised learning, which does not rely on any auxiliary sample and utilizes the endogenous correlation information among input samples. Besides, we propose Multi-pretext Attention Network (MAN), which exploits a specific attention mechanism to combine the traditional augmentation-relied methods and our GC, adaptively learning their optimized weights to improve the performance and enabling the feature extractor to obtain more universal representations. We evaluate our MAN extensively on miniImageNet and tieredImageNet datasets and the results demonstrate that the proposed method outperforms the state-of-the-art (SOTA) relevant methods.

cs.CV↗

arXiv2021

Over-sampling De-occlusion Attention Network for Prohibited Items Detection in Noisy X-ray Images

Security inspection is X-ray scanning for personal belongings in suitcases, which is significantly important for the public security but highly time-consuming for human inspectors. Fortunately, deep learning has greatly promoted the development of computer vision, offering a possible way of automatic security inspection. However, items within a luggage are randomly overlapped resulting in noisy X-ray images with heavy occlusions. Thus, traditional CNN-based models trained through common image recognition datasets fail to achieve satisfactory performance in this scenario. To address these problems, we contribute the first high-quality prohibited X-ray object detection dataset named OPIXray, which contains 8885 X-ray images from 5 categories of the widely-occurred prohibited item ``cutters''. The images are gathered from an airport and these prohibited items are annotated manually by professional inspectors, which can be used as a benchmark for model training and further facilitate future research. To better improve occluded X-ray object detection, we further propose an over-sampling de-occlusion attention network (DOAM-O), which consists of a novel de-occlusion attention module and a new over-sampling training strategy. Specifically, our de-occlusion module, namely DOAM, simultaneously leverages the different appearance information of the prohibited items; the over-sampling training strategy forces the model to put more emphasis on these hard samples consisting these items of high occlusion levels, which is more suitable for this scenario. We comprehensively evaluated DOAM-O on the OPIXray dataset, which proves that our model can stably improve the performance of the famous detection models such as SSD, YOLOv3, and FCOS, and outperform many extensively-used attention mechanisms.

cs.CV↗

arXiv2020

Predictive Control of Rural Microgrids with Temperature-dependent Battery Degradation Cost

Off-grid systems have emerged as a sustainable and cost-effective solution for rural electrification. In sub-Sarahan Africa (SSA), a great number of solar-hybrid microgrids have been installed or planned, operating stand-alone or grid-tied to a weak grid. Presence of intermittent energy sources necessitates the provision of energy storage for system balancing. Reliability and economic performance of those rural microgrids strongly depend on specific control strategies. This work develops a predictive control framework dedicated to rural microgrids incorporating a temperature-dependent battery degradation model. Based on a scalable DC PV-battery microgrid, the realistic simulation shows its superior performance in the reliability improvement and cost reduction. Compared with the day-ahead control without the temperature-dependent battery degradation model, this control strategy can improve the reliability by 5.5% and extend the lead-acid battery life time by 26%, equivalent to lowering the levelised cost of electricity (LCOE) by 13%.

eess.SY↗

arXiv2019

Assessment of the Value of Frequency Response Times in Power Systems

Given the increasing penetration in renewable generation, the UK power system is experiencing a decline in system inertia and an increase in frequency response (FR) requirements. Faster FR products are a mitigating solution that can cost-effectively meet the system balancing requirements. Thus, this paper proposes a mixed integer linear programming (MILP) unit commitment model which can simultaneously schedule inertial response, mandatory FR, as well as a sub-second FR product - enhanced frequency response (EFR). The model quantifies the value of providing faster reacting FR products in comparison with other response times from typical FR products. The performance and value of EFR are determined in a series of future energy scenarios with respect to the UK market and system conditions.

eess.SY↗