arXiv subjects

Ke Li

Publications and source records attributed to Ke Li.

At least 235 records · Page 13Linked to original sources

arXiv2024

IceFormer: Accelerated Inference with Long-Sequence Transformers on CPUs

One limitation of existing Transformer-based models is that they cannot handle very long sequences as input since their self-attention operations exhibit quadratic time and space complexity. This problem becomes especially acute when Transformers are deployed on hardware platforms equipped only with CPUs. To address this issue, we propose a novel method for accelerating self-attention at inference time that works with pretrained Transformer models out-of-the-box without requiring retraining. We experiment using our method to accelerate various long-sequence Transformers, including a leading LLaMA 2-based LLM, on various benchmarks and demonstrate a greater speedup of 2.73x - 7.63x while retaining 98.6% - 99.6% of the accuracy of the original pretrained models. The code is available on our project website at https://yuzhenmao.github.io/IceFormer/.

cs.LG↗

arXiv2024

Threshold and frequency properties of a cold ytterbium laser

We investigate properties of the lasing action observed on the 1S0--3P1 intercombination transition of ytterbium atoms that are laser-cooled and -trapped inside a high-finesse cavity. The dressing of the atomic states on the 1S0--1P1 transition by the magneto-optical trap (MOT) laser light allows the coupled atom-cavity system to lase, via a two-photon transition, on the same line on which it is pumped. The observation and basic description of this phenomenon was presented earlier by Gothe et al. [Phys. Rev. A 99, 013415 (2019)]. In the current work, we focus on a detailed analysis of the lasing threshold and frequency properties and perform a comparison to our theoretical models.

quant-ph↗

arXiv2024

AudioChatLlama: Towards General-Purpose Speech Abilities for LLMs

In this work, we extend the instruction-tuned Llama-2 model with end-to-end general-purpose speech processing and reasoning abilities while maintaining the wide range of original LLM capabilities, without using any carefully curated paired data. The resulting end-to-end model, named AudioChatLlama, can utilize audio prompts as a replacement for text and sustain a conversation. Such a model also has extended cross-modal capabilities such as being able to perform spoken question answering (QA), speech translation, and audio summarization amongst many other closed and open-domain tasks. This is unlike prior approaches in speech, in which LLMs are extended to handle audio for a limited number of pre-designated tasks. On both synthesized and recorded speech QA test sets, evaluations show that our end-to-end approach is on par with or outperforms cascaded systems (speech recognizer + LLM) in terms of modeling the response to a prompt. Furthermore, unlike cascades, our approach can interchange text and audio modalities and intrinsically utilize prior context in a conversation to provide better results.

cs.CL↗

arXiv2024

Xiwu: A Basis Flexible and Learnable LLM for High Energy Physics

Large Language Models (LLMs) are undergoing a period of rapid updates and changes, with state-of-the-art (SOTA) model frequently being replaced. When applying LLMs to a specific scientific field, it's challenging to acquire unique domain knowledge while keeping the model itself advanced. To address this challenge, a sophisticated large language model system named as Xiwu has been developed, allowing you switch between the most advanced foundation models and quickly teach the model domain knowledge. In this work, we will report on the best practices for applying LLMs in the field of high-energy physics (HEP), including: a seed fission technology is proposed and some data collection and cleaning tools are developed to quickly obtain domain AI-Ready dataset; a just-in-time learning system is implemented based on the vector store technology; an on-the-fly fine-tuning system has been developed to facilitate rapid training under a specified foundation model. The results show that Xiwu can smoothly switch between foundation models such as LLaMA, Vicuna, ChatGLM and Grok-1. The trained Xiwu model is significantly outperformed the benchmark model on the HEP knowledge question-and-answering and code generation. This strategy significantly enhances the potential for growth of our model's performance, with the hope of surpassing GPT-4 as it evolves with the development of open-source models. This work provides a customized LLM for the field of HEP, while also offering references for applying LLM to other fields, the corresponding codes are available on Github.

hep-ph↗

arXiv2024

LSA: Modeling Aspect Sentiment Coherency via Local Sentiment Aggregation

Aspect sentiment coherency is an intriguing yet underexplored topic in the field of aspect-based sentiment classification. This concept reflects the common pattern where adjacent aspects often share similar sentiments. Despite its prevalence, current studies have not fully recognized the potential of modeling aspect sentiment coherency, including its implications in adversarial defense. To model aspect sentiment coherency, we propose a novel local sentiment aggregation (LSA) paradigm based on constructing a differential-weighted sentiment aggregation window. We have rigorously evaluated our model through experiments, and the results affirm the proficiency of LSA in terms of aspect coherency prediction and aspect sentiment classification. For instance, it outperforms existing models and achieves state-of-the-art sentiment classification performance across five public datasets. Furthermore, we demonstrate the promising ability of LSA in ABSC adversarial defense, thanks to its sentiment coherency modeling. To encourage further exploration and application of this concept, we have made our code publicly accessible. This will provide researchers with a valuable tool to delve into sentiment coherency modeling in future research.

cs.CL↗

arXiv2024

BootAug: Boosting Text Augmentation via Hybrid Instance Filtering Framework

Text augmentation is an effective technique for addressing the problem of insufficient data in natural language processing. However, existing text augmentation methods tend to focus on few-shot scenarios and usually perform poorly on large public datasets. Our research indicates that existing augmentation methods often generate instances with shifted feature spaces, which leads to a drop in performance on the augmented data (for example, EDA generally loses $\approx 2\%$ in aspect-based sentiment classification). To address this problem, we propose a hybrid instance-filtering framework (BootAug) based on pre-trained language models that can maintain a similar feature space with natural datasets. BootAug is transferable to existing text augmentation methods (such as synonym substitution and back translation) and significantly improves the augmentation performance by $\approx 2-3\%$ in classification accuracy. Our experimental results on three classification tasks and nine public datasets show that BootAug addresses the performance drop problem and outperforms state-of-the-art text augmentation methods. Additionally, we release the code to help improve existing augmentation methods on large datasets.

cs.CL↗

arXiv2024

The Best Defense is Attack: Repairing Semantics in Textual Adversarial Examples

Recent studies have revealed the vulnerability of pre-trained language models to adversarial attacks. Existing adversarial defense techniques attempt to reconstruct adversarial examples within feature or text spaces. However, these methods struggle to effectively repair the semantics in adversarial examples, resulting in unsatisfactory performance and limiting their practical utility. To repair the semantics in adversarial examples, we introduce a novel approach named Reactive Perturbation Defocusing (Rapid). Rapid employs an adversarial detector to identify fake labels of adversarial examples and leverage adversarial attackers to repair the semantics in adversarial examples. Our extensive experimental results conducted on four public datasets, convincingly demonstrate the effectiveness of Rapid in various adversarial attack scenarios. To address the problem of defense performance validation in previous works, we provide a demonstration of adversarial detection and repair based on our work, which can be easily evaluated at https://tinyurl.com/22ercuf8.

cs.CL↗

arXiv2024

A General and Efficient Training for Transformer via Token Expansion

The remarkable performance of Vision Transformers (ViTs) typically requires an extremely large training cost. Existing methods have attempted to accelerate the training of ViTs, yet typically disregard method universality with accuracy dropping. Meanwhile, they break the training consistency of the original transformers, including the consistency of hyper-parameters, architecture, and strategy, which prevents them from being widely applied to different Transformer networks. In this paper, we propose a novel token growth scheme Token Expansion (termed ToE) to achieve consistent training acceleration for ViTs. We introduce an "initialization-expansion-merging" pipeline to maintain the integrity of the intermediate feature distribution of original transformers, preventing the loss of crucial learnable information in the training process. ToE can not only be seamlessly integrated into the training and fine-tuning process of transformers (e.g., DeiT and LV-ViT), but also effective for efficient training frameworks (e.g., EfficientTrain), without twisting the original training hyper-parameters, architecture, and introducing additional training strategies. Extensive experiments demonstrate that ToE achieves about 1.3x faster for the training of ViTs in a lossless manner, or even with performance gains over the full-token training baselines. Code is available at https://github.com/Osilly/TokenExpansion .

cs.LG↗

arXiv2024

EchoWrist: Continuous Hand Pose Tracking and Hand-Object Interaction Recognition Using Low-Power Active Acoustic Sensing On a Wristband

Our hands serve as a fundamental means of interaction with the world around us. Therefore, understanding hand poses and interaction context is critical for human-computer interaction. We present EchoWrist, a low-power wristband that continuously estimates 3D hand pose and recognizes hand-object interactions using active acoustic sensing. EchoWrist is equipped with two speakers emitting inaudible sound waves toward the hand. These sound waves interact with the hand and its surroundings through reflections and diffractions, carrying rich information about the hand's shape and the objects it interacts with. The information captured by the two microphones goes through a deep learning inference system that recovers hand poses and identifies various everyday hand activities. Results from the two 12-participant user studies show that EchoWrist is effective and efficient at tracking 3D hand poses and recognizing hand-object interactions. Operating at 57.9mW, EchoWrist is able to continuously reconstruct 20 3D hand joints with MJEDE of 4.81mm and recognize 12 naturalistic hand-object interactions with 97.6% accuracy.

cs.HC↗

arXiv2024

Task-Adaptive Saliency Guidance for Exemplar-free Class Incremental Learning

Exemplar-free Class Incremental Learning (EFCIL) aims to sequentially learn tasks with access only to data from the current one. EFCIL is of interest because it mitigates concerns about privacy and long-term storage of data, while at the same time alleviating the problem of catastrophic forgetting in incremental learning. In this work, we introduce task-adaptive saliency for EFCIL and propose a new framework, which we call Task-Adaptive Saliency Supervision (TASS), for mitigating the negative effects of saliency drift between different tasks. We first apply boundary-guided saliency to maintain task adaptivity and \textit{plasticity} on model attention. Besides, we introduce task-agnostic low-level signals as auxiliary supervision to increase the \textit{stability} of model attention. Finally, we introduce a module for injecting and recovering saliency noise to increase the robustness of saliency preservation. Our experiments demonstrate that our method can better preserve saliency maps across tasks and achieve state-of-the-art results on the CIFAR-100, Tiny-ImageNet, and ImageNet-Subset EFCIL benchmarks. Code is available at \url{https://github.com/scok30/tass}.

cs.CV↗

arXiv2024

Genetic Auto-prompt Learning for Pre-trained Code Intelligence Language Models

As Pre-trained Language Models (PLMs), a popular approach for code intelligence, continue to grow in size, the computational cost of their usage has become prohibitively expensive. Prompt learning, a recent development in the field of natural language processing, emerges as a potential solution to address this challenge. In this paper, we investigate the effectiveness of prompt learning in code intelligence tasks. We unveil its reliance on manually designed prompts, which often require significant human effort and expertise. Moreover, we discover existing automatic prompt design methods are very limited to code intelligence tasks due to factors including gradient dependence, high computational demands, and limited applicability. To effectively address both issues, we propose Genetic Auto Prompt (GenAP), which utilizes an elaborate genetic algorithm to automatically design prompts. With GenAP, non-experts can effortlessly generate superior prompts compared to meticulously manual-designed ones. GenAP operates without the need for gradients or additional computational costs, rendering it gradient-free and cost-effective. Moreover, GenAP supports both understanding and generation types of code intelligence tasks, exhibiting great applicability. We conduct GenAP on three popular code intelligence PLMs with three canonical code intelligence tasks including defect prediction, code summarization, and code translation. The results suggest that GenAP can effectively automate the process of designing prompts. Specifically, GenAP outperforms all other methods across all three tasks (e.g., improving accuracy by an average of 2.13% for defect prediction). To the best of our knowledge, GenAP is the first work to automatically design prompts for code intelligence PLMs.

cs.SE↗

arXiv2024

RESTORE: Towards Feature Shift for Vision-Language Prompt Learning

Prompt learning is effective for fine-tuning foundation models to improve their generalization across a variety of downstream tasks. However, the prompts that are independently optimized along a single modality path, may sacrifice the vision-language alignment of pre-trained models in return for improved performance on specific tasks and classes, leading to poorer generalization. In this paper, we first demonstrate that prompt tuning along only one single branch of CLIP (e.g., language or vision) is the reason why the misalignment occurs. Without proper regularization across the learnable parameters in different modalities, prompt learning violates the original pre-training constraints inherent in the two-tower architecture. To address such misalignment, we first propose feature shift, which is defined as the variation of embeddings after introducing the learned prompts, to serve as an explanatory tool. We dive into its relation with generalizability and thereafter propose RESTORE, a multi-modal prompt learning method that exerts explicit constraints on cross-modal consistency. To be more specific, to prevent feature misalignment, a feature shift consistency is introduced to synchronize inter-modal feature shifts by measuring and regularizing the magnitude of discrepancy during prompt tuning. In addition, we propose a "surgery" block to avoid short-cut hacking, where cross-modal misalignment can still be severe if the feature shift of each modality varies drastically at the same rate. It is implemented as feed-forward adapters upon both modalities to alleviate the misalignment problem. Extensive experiments on 15 datasets demonstrate that our method outperforms the state-of-the-art prompt tuning methods without compromising feature alignment.

cs.CV↗

arXiv2024

Sinkhorn Distance Minimization for Knowledge Distillation

Knowledge distillation (KD) has been widely adopted to compress large language models (LLMs). Existing KD methods investigate various divergence measures including the Kullback-Leibler (KL), reverse Kullback-Leibler (RKL), and Jensen-Shannon (JS) divergences. However, due to limitations inherent in their assumptions and definitions, these measures fail to deliver effective supervision when few distribution overlap exists between the teacher and the student. In this paper, we show that the aforementioned KL, RKL, and JS divergences respectively suffer from issues of mode-averaging, mode-collapsing, and mode-underestimation, which deteriorates logits-based KD for diverse NLP tasks. We propose the Sinkhorn Knowledge Distillation (SinKD) that exploits the Sinkhorn distance to ensure a nuanced and precise assessment of the disparity between teacher and student distributions. Besides, profit by properties of the Sinkhorn metric, we can get rid of sample-wise KD that restricts the perception of divergence in each teacher-student sample pair. Instead, we propose a batch-wise reformulation to capture geometric intricacies of distributions across samples in the high-dimensional space. Comprehensive evaluation on GLUE and SuperGLUE, in terms of comparability, validity, and generalizability, highlights our superiority over state-of-the-art methods on all kinds of LLMs with encoder-only, encoder-decoder, and decoder-only architectures.

cs.LG↗

arXiv2024

EyeEcho: Continuous and Low-power Facial Expression Tracking on Glasses

In this paper, we introduce EyeEcho, a minimally-obtrusive acoustic sensing system designed to enable glasses to continuously monitor facial expressions. It utilizes two pairs of speakers and microphones mounted on glasses, to emit encoded inaudible acoustic signals directed towards the face, capturing subtle skin deformations associated with facial expressions. The reflected signals are processed through a customized machine-learning pipeline to estimate full facial movements. EyeEcho samples at 83.3 Hz with a relatively low power consumption of 167 mW. Our user study involving 12 participants demonstrates that, with just four minutes of training data, EyeEcho achieves highly accurate tracking performance across different real-world scenarios, including sitting, walking, and after remounting the devices. Additionally, a semi-in-the-wild study involving 10 participants further validates EyeEcho's performance in naturalistic scenarios while participants engage in various daily activities. Finally, we showcase EyeEcho's potential to be deployed on a commercial-off-the-shelf (COTS) smartphone, offering real-time facial expression tracking.

cs.HC↗

arXiv2024

GazeTrak: Exploring Acoustic-based Eye Tracking on a Glass Frame

In this paper, we present GazeTrak, the first acoustic-based eye tracking system on glasses. Our system only needs one speaker and four microphones attached to each side of the glasses. These acoustic sensors capture the formations of the eyeballs and the surrounding areas by emitting encoded inaudible sound towards eyeballs and receiving the reflected signals. These reflected signals are further processed to calculate the echo profiles, which are fed to a customized deep learning pipeline to continuously infer the gaze position. In a user study with 20 participants, GazeTrak achieves an accuracy of 3.6° within the same remounting session and 4.9° across different sessions with a refreshing rate of 83.3 Hz and a power signature of 287.9 mW. Furthermore, we report the performance of our gaze tracking system fully implemented on an MCU with a low-power CNN accelerator (MAX78002). In this configuration, the system runs at up to 83.3 Hz and has a total power signature of 95.4 mW with a 30 Hz FPS.

cs.HC↗

arXiv2024

Multi-Fidelity Methods for Optimization: A Survey

Real-world black-box optimization often involves time-consuming or costly experiments and simulations. Multi-fidelity optimization (MFO) stands out as a cost-effective strategy that balances high-fidelity accuracy with computational efficiency through a hierarchical fidelity approach. This survey presents a systematic exploration of MFO, underpinned by a novel text mining framework based on a pre-trained language model. We delve deep into the foundational principles and methodologies of MFO, focusing on three core components -- multi-fidelity surrogate models, fidelity management strategies, and optimization techniques. Additionally, this survey highlights the diverse applications of MFO across several key domains, including machine learning, engineering design optimization, and scientific discovery, showcasing the adaptability and effectiveness of MFO in tackling complex computational challenges. Furthermore, we also envision several emerging challenges and prospects in the MFO landscape, spanning scalability, the composition of lower fidelities, and the integration of human-in-the-loop approaches at the algorithmic level. We also address critical issues related to benchmarking and the advancement of open science within the MFO community. Overall, this survey aims to catalyze further research and foster collaborations in MFO, setting the stage for future innovations and breakthroughs in the field.

cs.LG↗

arXiv2024

Unlocking the Secrets of Software Configuration Landscapes-Ruggedness, Accessibility, Escapability, and Transferability

Modern software systems are often highly configurable to tailor varied requirements from diverse stakeholders. Understanding the mapping between configurations and the desired performance attributes plays a fundamental role in advancing the controllability and tuning of the underlying system, yet has long been a dark hole of knowledge due to their black-box nature and the enormous combinatorial configuration space. In this paper, using $86$M evaluated configurations from three real-world systems on $32$ running workloads, we conducted one of its kind fitness landscape analysis (FLA) for configurable software systems. With comprehensive FLA methods, we for the first time show that: $i)$ the software configuration landscapes are fairly rugged, with numerous scattered local optima; $ii)$ nevertheless, the top local optima are highly accessible, featuring significantly larger basins of attraction; $iii)$ most inferior local optima are escapable with simple perturbations; $iv)$ landscapes of the same system with different workloads share structural similarities, which can be exploited to expedite heuristic search. Our results also provide valuable insights on the design of tailored meta-heuristics for configuration tuning; our FLA framework along with the collected data, build solid foundation for future research in this direction.

cs.SE↗

arXiv2024

Detecting highly collimated photon-jets from Higgs boson exotic decays with deep learning

Recently, there has been a growing focus on the search for anomalous objects beyond standard model (BSM) signatures at the Large Hadron Collider (LHC). This study investigates novel signatures involving highly collimated photons, referred to as photon-jets. These photon-jets can be generated from highly boosted BSM particles that decay into two or more collimated photons in the final state. Since these photons cannot be isolated from each other, they are treated as a single jet-like object rather than a multi-photon signature. The Higgs portal model is utilized as a prototype for studying photon-jet signatures. Specifically, GEANT4 is employed to simulate electromagnetic showers in an ATLAS-like electromagnetic calorimeter, and three machine learning techniques: Boosted Decision Trees (BDT), Convolutional Neural Networks (CNN), and Particle Flow Networks (PFN) are applied to effectively distinguish these photon-jet signatures from single photons and neutral pions within the SM backgrounds. Our models attain an identification efficiency exceeding $99\%$ for photon-jets, coupled with a rejection rate surpassing $99\%$ for SM backgrounds. Furthermore, the sensitivities for searching photon-jet signatures from the Higgs boson exotic decays at the High-Luminosity LHC are obtained.

hep-ph↗