PyTorchKR Papers

최신 논문

총 3,892편
태그·기간 등 상세 필터

카테고리 탐색

로그인 후 이 태그 구독

arXiv:2406.17051

Leveraging Knowledge Distillation for Lightweight Skin Cancer Classification: Balancing Accuracy and Computational Efficiency

Skin cancer is a major concern to public health, accounting for one-third of the reported cancers. If not detected early, the cancer has the potential for severe consequences. Recognizing the critical need for effective skin cancer classification, we address the limitations of existing models, which are often too large to deploy in areas with limited computational resources. In response, we present a knowledge distillation based approach for creating a lightweight yet high-performing classifier. The proposed solution involves fusing three models, namely ResNet152V2, ConvNeXtBase, and ViT Base, to create an effective teacher model. The teacher model is then employed to guide a lightweight student model of size 2.03 MB. This student model is further compressed to 469.77 KB using 16-bit quantization, enabling smooth incorporation into edge devices. With six-stage image preprocessing, data augmentation, and a rigorous ablation study, the model achieves an impressive accuracy of 98.75% on the HAM10000 dataset and 98.94% on the Kaggle dataset in classifying benign and malignant skin cancers. With its high accuracy and compact size, our model appears to be a potential choice for accurate skin cancer classification, particularly in resource-constrained settings.

arXiv:2406.17517

Preserving Node Distinctness in Graph Autoencoders via Similarity Distillation

Graph autoencoders (GAEs), as a kind of generative self-supervised learning approach, have shown great potential in recent years. GAEs typically rely on distance-based criteria, such as mean-square-error (MSE), to reconstruct the input graph. However, relying solely on a single reconstruction criterion may lead to a loss of distinctiveness in the reconstructed graph, causing nodes to collapse into similar representations and resulting in sub-optimal performance. To address this issue, we have developed a simple yet effective strategy to preserve the necessary distinctness in the reconstructed graph. Inspired by the knowledge distillation technique, we found that the dual encoder-decoder architecture of GAEs can be viewed as a teacher-student relationship. Therefore, we propose transferring the knowledge of distinctness from the raw graph to the reconstructed graph, achieved through a simple KL constraint. Specifically, we compute pairwise node similarity scores in the raw graph and reconstructed graph. During the training process, the KL constraint is optimized alongside the reconstruction criterion. We conducted extensive experiments across three types of graph tasks, demonstrating the effectiveness and generality of our strategy. This indicates that the proposed approach can be employed as a plug-and-play method to avoid vague reconstructions and enhance overall performance.

arXiv:2406.17328공식 코드 · Papers with Code

Dual-Space Knowledge Distillation for Large Language Models

Knowledge distillation (KD) is known as a promising solution to compress large language models (LLMs) via transferring their knowledge to smaller models. During this process, white-box KD methods usually minimize the distance between the output distributions of the two models so that more knowledge can be transferred. However, in the current white-box KD framework, the output distributions are from the respective output spaces of the two models, using their own prediction heads. We argue that the space discrepancy will lead to low similarity between the teacher model and the student model on both representation and distribution levels. Furthermore, this discrepancy also hinders the KD process between models with different vocabularies, which is common for current LLMs. To address these issues, we propose a dual-space knowledge distillation (DSKD) framework that unifies the output spaces of the two models for KD. On the basis of DSKD, we further develop a cross-model attention mechanism, which can automatically align the representations of the two models with different vocabularies. Thus, our framework is not only compatible with various distance functions for KD (e.g., KL divergence) like the current framework, but also supports KD between any two LLMs regardless of their vocabularies. Experiments on task-agnostic instruction-following benchmarks show that DSKD significantly outperforms the current white-box KD framework with various distance functions, and also surpasses existing KD methods for LLMs with different vocabularies.

arXiv:2406.17538공식 코드 · Papers with Code

Three-Stream Temporal-Shift Attention Network Based on Self-Knowledge Distillation for Micro-Expression Recognition

Micro-expressions are subtle facial movements that occur spontaneously when people try to conceal real emotions. Micro-expression recognition is crucial in many fields, including criminal analysis and psychotherapy. However, micro-expression recognition is challenging since micro-expressions have low intensity and public datasets are small in size. To this end, a three-stream temporal-shift attention network based on self-knowledge distillation called SKD-TSTSAN is proposed in this paper. Firstly, to address the low intensity of muscle movements, we utilize learning-based motion magnification modules to enhance the intensity of muscle movements. Secondly, we employ efficient channel attention modules in the local-spatial stream to make the network focus on facial regions that are highly relevant to micro-expressions. In addition, temporal shift modules are used in the dynamic-temporal stream, which enables temporal modeling with no additional parameters by mixing motion information from two different temporal domains. Furthermore, we introduce self-knowledge distillation into the micro-expression recognition task by introducing auxiliary classifiers and using the deepest section of the network for supervision, encouraging all blocks to fully explore the features of the training set. Finally, extensive experiments are conducted on four public datasets: CASME II, SAMM, MMEW, and CAS(ME)3. The experimental results demonstrate that our SKD-TSTSAN outperforms other existing methods and achieves new state-of-the-art performance. Our code will be available at https://github.com/GuanghaoZhu663/SKD-TSTSAN.

arXiv:2406.16524

The Privileged Students: On the Value of Initialization in Multilingual Knowledge Distillation

Knowledge distillation (KD) has proven to be a successful strategy to improve the performance of smaller models in many NLP tasks. However, most of the work in KD only explores monolingual scenarios. In this paper, we investigate the value of KD in multilingual settings. We find the significance of KD and model initialization by analyzing how well the student model acquires multilingual knowledge from the teacher model. Our proposed method emphasizes copying the teacher model's weights directly to the student model to enhance initialization. Our findings show that model initialization using copy-weight from the fine-tuned teacher contributes the most compared to the distillation process itself across various multilingual settings. Furthermore, we demonstrate that efficient weight initialization preserves multilingual capabilities even in low-resource scenarios.

arXiv:2406.15766

Continual Learning with Diffusion-based Generative Replay for Industrial Streaming Data

The Industrial Internet of Things (IIoT) integrates interconnected sensors and devices to support industrial applications, but its dynamic environments pose challenges related to data drift. Considering the limited resources and the need to effectively adapt models to new data distributions, this paper introduces a Continual Learning (CL) approach, i.e., Distillation-based Self-Guidance (DSG), to address challenges presented by industrial streaming data via a novel generative replay mechanism. DSG utilizes knowledge distillation to transfer knowledge from the previous diffusion-based generator to the updated one, improving both the stability of the generator and the quality of reproduced data, thereby enhancing the mitigation of catastrophic forgetting. Experimental results on CWRU, DSA, and WISDM datasets demonstrate the effectiveness of DSG. DSG outperforms the state-of-the-art baseline in accuracy, demonstrating improvements ranging from 2.9% to 5.0% on key datasets, showcasing its potential for practical industrial applications.

arXiv:2406.16525공식 코드 · Papers with Code

Enhancing OOD Detection Using Latent Diffusion

Numerous Out-of-Distribution (OOD) detection algorithms have been developed to identify unknown samples or objects in real-world deployments. One line of work related to OOD detection propose utilizing auxiliary datasets to train OOD detectors, thereby enhancing the performance of OOD detection. Recently, researchers propose to leverage Stable Diffusion (SD) to generate outliers in the pixel space, which may complicate network training. To mitigate this issue, we propose an Outlier Aware Learning (OAL) framework, which synthesizes OOD training data in the latent space. This improvement enables us to train the network with only a few synthesized outliers. Besides, to regularize the model's decision boundary, we develop a mutual information-based contrastive learning module (MICL) that amplifies the distinction between In-Distribution (ID) and collected OOD features. Moreover, we develop a knowledge distillation module to prevent the degradation of ID classification accuracy when training with OOD data. Extensive experiments on CIFAR-10/100 benchmarks demonstrate the superior performance of our method.

arXiv:2406.14709

Factual Dialogue Summarization via Learning from Large Language Models

Factual consistency is an important quality in dialogue summarization. Large language model (LLM)-based automatic text summarization models generate more factually consistent summaries compared to those by smaller pretrained language models, but they face deployment challenges in real-world applications due to privacy or resource constraints. In this paper, we investigate the use of symbolic knowledge distillation to improve the factual consistency of smaller pretrained models for dialogue summarization. We employ zero-shot learning to extract symbolic knowledge from LLMs, generating both factually consistent (positive) and inconsistent (negative) summaries. We then apply two contrastive learning objectives on these summaries to enhance smaller summarization models. Experiments with BART, PEGASUS, and Flan-T5 indicate that our approach surpasses strong baselines that rely on complex data augmentation strategies. Our approach achieves better factual consistency while maintaining coherence, fluency, and relevance, as confirmed by various automatic evaluation metrics. We also provide access to the data and code to facilitate future research.

arXiv:2406.14847

Fair Text to Medical Image Diffusion Model with Subgroup Distribution Aligned Tuning

The text to medical image (T2MedI) with latent diffusion model has great potential to alleviate the scarcity of medical imaging data and explore the underlying appearance distribution of lesions in a specific patient status description. However, as the text to nature image models, we show that the T2MedI model can also bias to some subgroups to overlook the minority ones in the training set. In this work, we first build a T2MedI model based on the pre-trained Imagen model, which has the fixed contrastive language-image pre-training (CLIP) text encoder, while its decoder has been fine-tuned on medical images from the Radiology Objects in COntext (ROCO) dataset. Its gender bias is analyzed qualitatively and quantitatively. Toward this issue, we propose to fine-tune the T2MedI toward the target application dataset to align their sensitive subgroups distribution probability. Specifically, the alignment loss for fine-tuning is guided by an off-the-shelf sensitivity-subgroup classifier to match the classification probability between the generated images and the expected target dataset. In addition, the image quality is maintained by a CLIP-consistency regularization term following a knowledge distillation scheme. For evaluation, we set the target dataset to be enhanced as the BraST18 dataset, and trained a brain magnetic resonance (MR) slice-based gender classifier from it. With our method, the generated MR image can markedly reduce the inconsistency with the gender proportion in the BraTS18 dataset.

arXiv:2406.14629공식 코드 · Papers with Code

Can LLMs Learn by Teaching for Better Reasoning? A Preliminary Study

Teaching to improve student models (e.g., knowledge distillation) is an extensively studied methodology in LLMs. However, for humans, teaching improves not only students but also teachers, by fostering more rigorous and clear reasoning as well as knowledge building. We ask: Can LLMs also learn by teaching (LbT) for better reasoning? If the answer is yes, we can potentially unlock the possibility of continuously advancing the models without solely relying on human-produced data or stronger models. In this paper, we provide a preliminary exploration on this question. We show that LbT ideas can be incorporated into existing LLM training/prompting pipelines and bring improvements. Specifically, we design three methods, each mimicking one of the three levels of LbT: observing students' feedback, learning from the feedback, and learning iteratively, with the goals of improving answer accuracy without training or improving models' inherent capability with fine-tuning. We reveal some findings: (1) Teaching materials that make it easier for students to learn have clearer and more accurate logic when using in-context learning as the student's "learning" method; (2) Weak-to-strong generalization: LbT might help improve strong models by teaching weak models; (3) Diversity in students might help: teaching multiple students could be better than teaching one student or the teacher itself. We hope that our exploration can inspire future research on LbT and more broadly adopting the advanced techniques in education to improve LLMs. The code and website are at https://github.com/imagination-research/lbt and https://sites.google.com/view/llm-learning-by-teaching.

arXiv:2406.13808

Can Low-Rank Knowledge Distillation in LLMs be Useful for Microelectronic Reasoning?

In this work, we present empirical results regarding the feasibility of using offline large language models (LLMs) in the context of electronic design automation (EDA). The goal is to investigate and evaluate a contemporary language model's (Llama-2-7B) ability to function as a microelectronic Q & A expert as well as its reasoning, and generation capabilities in solving microelectronic-related problems. Llama-2-7B was tested across a variety of adaptation methods, including introducing a novel low-rank knowledge distillation (LoRA-KD) scheme. Our experiments produce both qualitative and quantitative results.

arXiv:2406.13344공식 코드 · Papers with Code

WaterMono: Teacher-Guided Anomaly Masking and Enhancement Boosting for Robust Underwater Self-Supervised Monocular Depth Estimation

Depth information serves as a crucial prerequisite for various visual tasks, whether on land or underwater. Recently, self-supervised methods have achieved remarkable performance on several terrestrial benchmarks despite the absence of depth annotations. However, in more challenging underwater scenarios, they encounter numerous brand-new obstacles such as the influence of marine life and degradation of underwater images, which break the assumption of a static scene and bring low-quality images, respectively. Besides, the camera angles of underwater images are more diverse. Fortunately, we have discovered that knowledge distillation presents a promising approach for tackling these challenges. In this paper, we propose WaterMono, a novel framework for depth estimation coupled with image enhancement. It incorporates the following key measures: (1) We present a Teacher-Guided Anomaly Mask to identify dynamic regions within the images; (2) We employ depth information combined with the Underwater Image Formation Model to generate enhanced images, which in turn contribute to the depth estimation task; and (3) We utilize a rotated distillation strategy to enhance the model's rotational robustness. Comprehensive experiments demonstrate the effectiveness of our proposed method for both depth estimation and image enhancement. The source code and pre-trained models are available on the project home page: https://github.com/OUCVisionGroup/WaterMono.

arXiv:2406.13114공식 코드 · Papers with Code

Multi-Stage Balanced Distillation: Addressing Long-Tail Challenges in Sequence-Level Knowledge Distillation

Large language models (LLMs) have significantly advanced various natural language processing tasks, but deploying them remains computationally expensive. Knowledge distillation (KD) is a promising solution, enabling the transfer of capabilities from larger teacher LLMs to more compact student models. Particularly, sequence-level KD, which distills rationale-based reasoning processes instead of merely final outcomes, shows great potential in enhancing students' reasoning capabilities. However, current methods struggle with sequence level KD under long-tailed data distributions, adversely affecting generalization on sparsely represented domains. We introduce the Multi-Stage Balanced Distillation (BalDistill) framework, which iteratively balances training data within a fixed computational budget. By dynamically selecting representative head domain examples and synthesizing tail domain examples, BalDistill achieves state-of-the-art performance across diverse long-tailed datasets, enhancing both the efficiency and efficacy of the distilled models.

arXiv:2406.13555공식 코드 · Papers with Code

BiLD: Bi-directional Logits Difference Loss for Large Language Model Distillation

In recent years, large language models (LLMs) have shown exceptional capabilities across various natural language processing (NLP) tasks. However, such impressive performance often comes with the trade-off of an increased parameter size, posing significant challenges for widespread deployment. Knowledge distillation (KD) provides a solution by transferring knowledge from a large teacher model to a smaller student model. In this paper, we explore the task-specific distillation of LLMs at the logit level. Our investigation reveals that the logits of fine-tuned LLMs exhibit a more extreme long-tail distribution than those from vision models, with hidden "noise" in the long tail affecting distillation performance. Furthermore, existing logits distillation methods often struggle to effectively utilize the internal ranking information from the logits. To address these, we propose the Bi-directional Logits Difference (BiLD) loss. The BiLD loss filters out the long-tail noise by utilizing only top-$k$ teacher and student logits, and leverages the internal logits ranking information by constructing logits differences. To evaluate BiLD loss, we conduct comprehensive experiments on 13 datasets using two types of LLMs. Our results show that the BiLD loss, with only the top-8 logits, outperforms supervised fine-tuning (SFT), vanilla KL loss, and five other distillation methods from both NLP and CV fields.

arXiv:2406.14282공식 코드 · Papers with Code

Learning to Plan for Retrieval-Augmented Large Language Models from Knowledge Graphs

Improving the performance of large language models (LLMs) in complex question-answering (QA) scenarios has always been a research focal point. Recent studies have attempted to enhance LLMs' performance by combining step-wise planning with external retrieval. While effective for advanced models like GPT-3.5, smaller LLMs face challenges in decomposing complex questions, necessitating supervised fine-tuning. Previous work has relied on manual annotation and knowledge distillation from teacher LLMs, which are time-consuming and not accurate enough. In this paper, we introduce a novel framework for enhancing LLMs' planning capabilities by using planning data derived from knowledge graphs (KGs). LLMs fine-tuned with this data have improved planning capabilities, better equipping them to handle complex QA tasks that involve retrieval. Evaluations on multiple datasets, including our newly proposed benchmark, highlight the effectiveness of our framework and the benefits of KG-derived planning data.

arXiv:2406.14208

SeCoKD: Aligning Large Language Models for In-Context Learning with Fewer Shots

Previous studies have shown that demonstrations can significantly help Large Language Models (LLMs ) perform better on the given tasks. However, this so-called In-Context Learning ( ICL ) ability is very sensitive to the presenting context, and often dozens of demonstrations are needed. In this work, we investigate if we can reduce the shot number while still maintaining a competitive performance. We present SeCoKD, a self-Knowledge Distillation ( KD ) training framework that aligns the student model with a heavily prompted variation, thereby increasing the utilization of a single demonstration. We experiment with the SeCoKD across three LLMs and six benchmarks focusing mainly on reasoning tasks. Results show that our method outperforms the base model and Supervised Fine-tuning ( SFT ), especially in zero-shot and one-shot settings by 30% and 10%, respectively. Moreover, SeCoKD brings little negative artifacts when evaluated on new tasks, which is more robust than Supervised Fine-tuning.

arXiv:2406.14185

Failure-Resilient Distributed Inference with Model Compression over Heterogeneous Edge Devices

The distributed inference paradigm enables the computation workload to be distributed across multiple devices, facilitating the implementations of deep learning based intelligent services on extremely resource-constrained Internet of Things (IoT) scenarios. Yet it raises great challenges to perform complicated inference tasks relying on a cluster of IoT devices that are heterogeneous in their computing/communication capacity and prone to crash or timeout failures. In this paper, we present RoCoIn, a robust cooperative inference mechanism for locally distributed execution of deep neural network-based inference tasks over heterogeneous edge devices. It creates a set of independent and compact student models that are learned from a large model using knowledge distillation for distributed deployment. In particular, the devices are strategically grouped to redundantly deploy and execute the same student model such that the inference process is resilient to any local failures, while a joint knowledge partition and student model assignment scheme are designed to minimize the response latency of the distributed inference system in the presence of devices with diverse capacities. Extensive simulations are conducted to corroborate the superior performance of our RoCoIn for distributed inference compared to several baselines, and the results demonstrate its efficacy in timely inference and failure resiliency.

논문 ID: document/1046864

eDifFIQA: Towards Efficient Face Image Quality Assessment Based On Denoising Diffusion Probabilistic Models

State-of-the-art Face Recognition (FR) models perform well in constrained scenarios, but frequently fail in difficult real-world scenarios, when no quality guarantees can be made for face samples. For this reason, Face Image Quality Assessment (FIQA) techniques are often used by FR systems, to provide quality estimates of captured face samples. The quality estimate provided by FIQA techniques can be used by the FR system to reject samples of low-quality, in turn improving the performance of the system and reducing the number of critical false-match errors. However, despite steady improvements, ensuring a good trade-off between the performance and computational complexity of FIQA methods across diverse face samples remains challenging. In this paper, we present DifFIQA, a powerful unsupervised approach for quality assessment based on the popular denoising diffusion probabilistic models (DDPMs) and the extended (eDifFIQA) approach. The main idea of the base DifFIQA approach is to utilize the forward and backward processes of DDPMs to perturb facial images and quantify the impact of these perturbations on the corresponding image embeddings for quality prediction. Because of the iterative nature of DDPMs the base DifFIQA approach is extremely computationally expensive. Using eDifFIQA we are able to improve on both the performance and computational complexity of the base DifFIQA approach, by employing label optimized knowledge distillation. In this process, quality information inferred by DifFIQA is distilled into a quality-regression model. During the distillation process we use an additional source of quality information hidden in the relative position of the embedding to further improve the predictive capabilities of the underlying regression model. By choosing different feature extraction backbone models as the basis for the quality-regression eDifFIQA model, we are able to control the trade-off between the predictive capabilities and computational complexity of the final model. We evaluate three eDifFIQA variants of varying sizes in comprehensive experiments on 7 diverse datasets containing static-images and a separate video-based dataset, with 4 target CNN-based FR models and 2 target Transformer-based FR models and against 10 state-of-the-art FIQA techniques, as well as against the initial DifFIQA baseline and a simple regression-based predictor DifFIQA(R), distilled from DifFIQA without any additional optimization. The results show that the proposed label optimized knowledge distillation improves on the performance and computationally complexity of the base DifFIQA approach, and is able to achieve state-of-the-art performance in several distinct experimental scenarios. Furthermore, we also show that the distilled model can be used directly for face recognition and leads to highly competitive results.

arXiv:2406.12169

Intermediate Distillation: Data-Efficient Distillation from Black-Box LLMs for Information Retrieval

Recent research has explored distilling knowledge from large language models (LLMs) to optimize retriever models, especially within the retrieval-augmented generation (RAG) framework. However, most existing training methods rely on extracting supervision signals from LLMs' weights or their output probabilities, which is not only resource-intensive but also incompatible with black-box LLMs. In this paper, we introduce \textit{Intermediate Distillation}, a data-efficient knowledge distillation training scheme that treats LLMs as black boxes and distills their knowledge via an innovative LLM-ranker-retriever pipeline, solely using LLMs' ranking generation as the supervision signal. Extensive experiments demonstrate that our proposed method can significantly improve the performance of retriever models with only 1,000 training instances. Moreover, our distilled retriever model significantly boosts performance in question-answering tasks within the RAG framework, demonstrating the potential of LLMs to economically and effectively train smaller models.

arXiv:2406.09444

GenDistiller: Distilling Pre-trained Language Models based on an Autoregressive Generative Model

Pre-trained speech language models such as HuBERT and WavLM leverage unlabeled speech data for self-supervised learning and offer powerful representations for numerous downstream tasks. Despite the success of these models, their high requirements for memory and computing resource hinder their application on resource restricted devices. Therefore, this paper introduces GenDistiller, a novel knowledge distillation framework which generates the hidden representations of the pre-trained teacher model directly by a much smaller student network. The proposed method takes the previous hidden layer as history and implements a layer-by-layer prediction of the teacher model autoregressively. Experiments on SUPERB reveal the advantage of GenDistiller over the baseline distilling method without an autoregressive framework, with 33% fewer parameters, similar time consumption and better performance on most of the SUPERB tasks. Ultimately, the proposed GenDistiller reduces the size of WavLM by 82%.