Knowledge distillation is an effective approach to leverage a well-trained network or an ensemble of them, named as the teacher, to guide the training of a student network. The outputs from the teacher network are used as soft labels for supervising the training of a new network. Recent studies \citep{muller2019does,yuan2020revisiting} revealed an intriguing property of the soft labels that making labels soft serves as a good regularization to the student network. From the perspective of statistical learning, regularization aims to reduce the variance, however how bias and variance change is not clear for training with soft labels. In this paper, we investigate the bias-variance tradeoff brought by distillation with soft labels. Specifically, we observe that during training the bias-variance tradeoff varies sample-wisely. Further, under the same distillation temperature setting, we observe that the distillation performance is negatively associated with the number of some specific samples, which are named as regularization samples since these samples lead to bias increasing and variance decreasing. Nevertheless, we empirically find that completely filtering out regularization samples also deteriorates distillation performance. Our discoveries inspired us to propose the novel weighted soft labels to help the network adaptively handle the sample-wise bias-variance tradeoff. Experiments on standard evaluation benchmarks validate the effectiveness of our method. Our code is available at \url{https://github.com/bellymonster/Weighted-Soft-Label-Distillation}.
Rethinking Data Distillation: Do Not Overlook Calibration
Texas A&M University（德州农工大学）；Zhejiang University（浙江大学）；Certik；University of Maryland, College Park（马里兰大学帕克分校）；Purdue University（普渡大学）；North Carolina State University（北卡罗来纳州立大学）
Recent researches on unsupervised domain adaptation (UDA) have demonstrated that end-to-end ensemble learning frameworks serve as a compelling option for UDA tasks. Nevertheless, these end-to-end ensemble learning methods often lack flexibility as any modification to the ensemble requires retraining of their frameworks. To address this problem, we propose a flexible ensemble-distillation framework for performing semantic segmentation based UDA, allowing any arbitrary composition of the members in the ensemble while still maintaining its superior performance. To achieve such flexibility, our fr…
School of Computer Science, The University of Sydney（悉尼大学计算机科学学院）
Recent years have witnessed dramatically improvements in the knowledge distillation, which can generate a compact student model for better efficiency while retaining the model effectiveness of the teacher model. Previous studies find that: more accurate teachers do not necessary make for better teachers due to the mismatch of abilities. In this paper, we aim to analysis the phenomenon from the perspective of model calibration. We found that the larger teacher model may be too over-confident, thus the student model cannot effectively imitate. While, after the simple model calibration of the tea…
Rethinking Decoupled Knowledge Distillation: A Predictive Distribution Perspective
Department of Data Science and Artificial Intelligence, The Hong Kong Polytechnic University（数据科学与人工智能系，香港理工大学）；Department of Computing, The Hong Kong Polytechnic University（计算系，香港理工大学）；The Hong Kong Polytechnic University Shenzhen Research Institute（香港理工大学深圳研究院）
Rethinking Reverse KL as Adaptive Entropy Distillation
School of Computer Science and Engineering, Sun Yat-sen University（中山大学计算机科学与工程学院）；School of Science and Technology, Hong Kong Metropolitan University（香港都会大学科技学院）
Rethinking Dataset Distillation: Hard Truths about Soft Labels
Vision and AI Lab, Indian Institute of Science, Bangalore（印度科学研究院视觉与人工智能实验室，班加罗尔）；University of Maryland, Baltimore County（马里兰大学巴尔的摩县分校）；IIT Hyderabad（海得拉巴理工学院）
Tsinghua Shenzhen International Graduate School（清华大学深圳国际研究生院）；International Digital Economy Academy (IDEA)（国际数字经济研究院（IDEA））
Knowledge Distillation (KD) has developed extensively and boosted various tasks. The classical KD method adds the KD loss to the original cross-entropy (CE) loss. We try to decompose the KD loss to explore its relation with the CE loss. Surprisingly, we find it can be regarded as a combination of the CE loss and an extra loss which has the identical form as the CE loss. However, we notice the extra loss forces the student's relative probability to learn the teacher's absolute probability. Moreover, the sum of the two probabilities is different, making it hard to optimize. To address this issue…
Click-through rate (CTR) Prediction is of great importance in real-world online ads systems. One challenge for the CTR prediction task is to capture the real interest of users from their clicked items, which is inherently biased by presented positions of items, i.e., more front positions tend to obtain higher CTR values. A popular line of existing works focuses on explicitly estimating position bias by result randomization which is expensive and inefficient, or by inverse propensity weighting (IPW) which relies heavily on the quality of the propensity estimation. Another common solution is mod…
Benefiting from masked visual modeling, self-supervised video representation learning has achieved remarkable progress. However, existing methods focus on learning representations from scratch through reconstructing low-level features like raw pixel RGB values. In this paper, we propose masked video distillation (MVD), a simple yet effective two-stage masked feature modeling framework for video representation learning: firstly we pretrain an image (or video) model by recovering low-level features of masked patches, then we use the resulting features as targets for masked feature modeling. For …
Google Research（谷歌研究院）；University of California, Berkeley（加州大学伯克利分校）；Cohere For AI（Cohere人工智能研究机构）
Knowledge distillation has proven to be an effective technique in improving the performance a student model using predictions from a teacher model. However, recent work has shown that gains in average efficiency are not uniform across subgroups in the data, and in particular can often come at the cost of accuracy on rare subgroups and classes. To preserve strong performance across classes that may follow a long-tailed distribution, we develop distillation techniques that are tailored to improve the student's worst-class performance. Specifically, we introduce robust optimization objectives in …
East China Normal University（华东师范大学）；Huawei Technologies（华为技术有限公司）；Institute of Automation, Chinese Academy of Sciences（中国科学院自动化研究所）；Xiamen University（厦门大学）；Shanghai Jiao Tong University（上海交通大学）；Shenzhen University（深圳大学）；University of Sydney（悉尼大学）
Information Bottleneck (IB) based multi-view learning provides an information theoretic principle for seeking shared information contained in heterogeneous data descriptions. However, its great success is generally attributed to estimate the multivariate mutual information which is intractable when the network becomes complicated. Moreover, the representation learning tradeoff, {\it i.e.}, prediction-compression and sufficiency-consistency tradeoff, makes the IB hard to satisfy both requirements simultaneously. In this paper, we design several variational information bottlenecks to exploit two…
Dataset Distillation via Adversarial Prediction Matching
The Hong Kong University of Science and Technology (Guangzhou)（香港科技大学（广州））；The Hong Kong University of Science and Technology（香港科技大学）；CSIRO Data61（澳大利亚联邦科学与工业研究组织数据61所）；University of Electronic Science and Technology of China（电子科技大学）；Dongguan University of Technology（东莞理工学院）
National University of Defense Technology（国防科技大学）；University of Oulu（奥卢大学）；Nanyang Technological University（南洋理工大学）；Stanford University（斯坦福大学）
Few-Shot Class-Incremental Learning (FSCIL) aims at incrementally learning novel classes from a few labeled samples by avoiding the overfitting and catastrophic forgetting simultaneously. The current protocol of FSCIL is built by mimicking the general class-incremental learning setting, while it is not totally appropriate due to the different data configuration, i.e., novel classes are all in the limited data regime. In this paper, we rethink the configuration of FSCIL with the open-set hypothesis by reserving the possibility in the first session for incoming categories. To assign better perfo…
Knowledge distillation is a technique for improving the performance of a simple "student" model by replacing its one-hot training labels with a distribution over labels obtained from a complex "teacher" model. While this simple approach has proven widely effective, a basic question remains unresolved: why does distillation help? In this paper, we present a statistical perspective on distillation which addresses this question, and provides a novel connection to extreme multiclass retrieval techniques. Our core observation is that the teacher seeks to estimate the underlying (Bayes) class-probab…
Recent advances in model compression have provided procedures for compressing large neural networks to a fraction of their original size while retaining most if not all of their accuracy. However, all of these approaches rely on access to the original training set, which might not always be possible if the network to be compressed was trained on a very large dataset, or on a dataset whose release poses privacy or safety concerns as may be the case for biometrics tasks. We present a method for data-free knowledge distillation, which is able to compress deep neural networks trained on large-scal…
The University of Hong Kong（香港大学）；ByteDance Inc.（字节跳动公司）
In this paper, we empirically study how to make the most of low-resolution frames for efficient video recognition. Existing methods mainly focus on developing compact networks or alleviating temporal redundancy of video inputs to increase efficiency, whereas compressing frame resolution has rarely been considered a promising solution. A major concern is the poor recognition accuracy on low-resolution frames. We thus start by analyzing the underlying causes of performance degradation on low-resolution frames. Our key finding is that the major cause of degradation is not information loss in the …
Most of the current state-of-the-art frameworks for cross-season visual place recognition (CS-VPR) focus on domain adaptation (DA) to a single specific season. From the viewpoint of long-term CS-VPR, such frameworks do not scale well to sequential multiple domains (e.g., spring - summer - autumn - winter - ... ). The goal of this study is to develop a novel long-term ensemble learning (LEL) framework that allows for a constant cost retraining in long-term sequential-multi-domain CS-VPR (SMD-VPR), which only requires the memorization of a small constant number of deep convolutional neural netwo…
Object detection has achieved promising performance on clean datasets, but how to achieve better tradeoff between the adversarial robustness and clean precision is still under-explored. Adversarial training is the mainstream method to improve robustness, but most of the works will sacrifice clean precision to gain robustness than standard training. In this paper, we propose Unified Decoupled Feature Alignment (UDFA), a novel fine-tuning paradigm which achieves better performance than existing methods, by fully exploring the combination between self-knowledge distillation and adversarial traini…
INRIA（法国国家信息与自动化研究所）；Indian Institute of Information Technology, Allahabad（阿拉哈巴德印度信息技术学院）；Université Côte d’Azur（蔚蓝海岸大学）
Multimodal Deep Learning has garnered much interest, and transformers have triggered novel approaches, thanks to the cross-attention mechanism. Here we propose an approach to deal with two key existing challenges: the high computational resource demanded and the issue of missing modalities. We introduce for the first time the concept of knowledge distillation in transformers to use only one modality at inference time. We report a full study analyzing multiple student-teacher configurations, levels at which distillation is applied, and different methodologies. With the best configuration, we im…
Istituto Italiano di Tecnologia（意大利理工学院）；University of Genova（热那亚大学）；Huawei Technologies Co. Ltd.（华为技术有限公司）；University of Verona（维罗纳大学）
The concept of compressing deep Convolutional Neural Networks (CNNs) is essential to use limited computation, power, and memory resources on embedded devices. However, existing methods achieve this objective at the cost of a drop in inference accuracy in computer vision tasks. To address such a drawback, we propose a framework that leverages knowledge distillation along with customizable block-wise optimization to learn a lightweight CNN structure while preserving better control over the compression-performance tradeoff. Considering specific resource constraints, e.g., floating-point operation…
Lossy compression algorithms are typically designed and analyzed through the lens of Shannon's rate-distortion theory, where the goal is to achieve the lowest possible distortion (e.g., low MSE or high SSIM) at any given bit rate. However, in recent years, it has become increasingly accepted that "low distortion" is not a synonym for "high perceptual quality", and in fact optimization of one often comes at the expense of the other. In light of this understanding, it is natural to seek for a generalization of rate-distortion theory which takes perceptual quality into account. In this paper, we …
Rethinking Output Alignment For 1-bit Post-Training Quantization of Large Language Models
Department of Data Science and AI（数据科学与人工智能系）；Monash University（墨尔本大学）；Centre for Vision, Speech and Signal Processing（视觉、语音和信号处理中心）；University of Surrey（萨里大学）
Simon Fraser University（西蒙菲莎大学）；Tianjin University（天津大学）；Huawei Canada Technologies Co. Ltd.（华为加拿大技术有限公司）
Image-to-image translation models are shown to be vulnerable to the Membership Inference Attack (MIA), in which the adversary's goal is to identify whether a sample is used to train the model or not. With daily increasing applications based on image-to-image translation models, it is crucial to protect the privacy of these models against MIAs. We propose adversarial knowledge distillation (AKD) as a defense method against MIAs for image-to-image translation models. The proposed method protects the privacy of the training samples by improving the generalizability of the model. We conduct experi…
Triple-View Knowledge Distillation for Semi-Supervised Semantic Segmentation
School of Computer Science and Technology, Hangzhou Dianzi University（杭州电子科技大学计算机科学与技术学院）；Guangdong Laboratory of Artificial Intelligence and Digital Economy (SZ)（广东人工智能与数字经济实验室）；School of Electronic and Computer Engineering, Peking University（北京大学电子与计算机工程学院）；PengCheng Laboratory（鹏城实验室）；College of Computer Science, Zhejiang University（浙江大学计算机学院）
Is Smaller Always Faster? Tradeoffs in Compressing Self-Supervised Speech Transformers
Graduate Institute of Communication Engineering, National Taiwan University, Taiwan（台湾国立台湾大学通信工程研究所）；University of California, Los Angeles, United States（美国加州大学洛杉矶分校）；University of Washington, United States（美国华盛顿大学）；University of Edinburgh, United Kingdom（英国爱丁堡大学）
