CKD: Contrastive Knowledge Distillation from A Sample-wise Perspective

• URL: http://arxiv.org/abs/2404.14109v1
• Date: Mon, 22 Apr 2024 11:52:40 GMT
• Title: CKD: Contrastive Knowledge Distillation from A Sample-wise Perspective
• Authors: Wencheng Zhu, Xin Zhou, Pengfei Zhu, Yu Wang, Qinghua Hu,
• Abstract summary: We present a contrastive knowledge distillation approach, which can be formulated as a sample-wise alignment problem with intra- and inter-sample constraints. Our method minimizes logit differences within the same sample by considering their numerical values. We conduct comprehensive experiments on three datasets including CIFAR-100, ImageNet-1K, and MS COCO.
• Score: 48.99488315273868
• License: http://creativecommons.org/licenses/by-sa/4.0/
• Abstract: In this paper, we present a simple yet effective contrastive knowledge distillation approach, which can be formulated as a sample-wise alignment problem with intra- and inter-sample constraints. Unlike traditional knowledge distillation methods that concentrate on maximizing feature similarities or preserving class-wise semantic correlations between teacher and student features, our method attempts to recover the "dark knowledge" by aligning sample-wise teacher and student logits. Specifically, our method first minimizes logit differences within the same sample by considering their numerical values, thus preserving intra-sample similarities. Next, we bridge semantic disparities by leveraging dissimilarities across different samples. Note that constraints on intra-sample similarities and inter-sample dissimilarities can be efficiently and effectively reformulated into a contrastive learning framework with newly designed positive and negative pairs. The positive pair consists of the teacher's and student's logits derived from an identical sample, while the negative pairs are formed by using logits from different samples. With this formulation, our method benefits from the simplicity and efficiency of contrastive learning through the optimization of InfoNCE, yielding a run-time complexity that is far less than $O(n^2)$, where $n$ represents the total number of training samples. Furthermore, our method can eliminate the need for hyperparameter tuning, particularly related to temperature parameters and large batch sizes. We conduct comprehensive experiments on three datasets including CIFAR-100, ImageNet-1K, and MS COCO. Experimental results clearly confirm the effectiveness of the proposed method on both image classification and object detection tasks. Our source codes will be publicly available at https://github.com/wencheng-zhu/CKD.

Related papers

• Task-oriented Embedding Counts: Heuristic Clustering-driven Feature Fine-tuning for Whole Slide Image Classification [1.292108130501585]
  We propose a clustering-driven feature fine-tuning method (HC-FT) to enhance the performance of multiple instance learning. The proposed method is evaluated on both CAMELYON16 and BRACS datasets, achieving an AUC of 97.13% and 85.85%, respectively.
  arXiv  Detail & Related papers  (2024-06-02T08:53:45Z)
• Noisy Correspondence Learning with Self-Reinforcing Errors Mitigation [63.180725016463974]
  Cross-modal retrieval relies on well-matched large-scale datasets that are laborious in practice. We introduce a novel noisy correspondence learning framework, namely textbfSelf-textbfReinforcing textbfErrors textbfMitigation (SREM)
  arXiv  Detail & Related papers  (2023-12-27T09:03:43Z)
• Additional Positive Enables Better Representation Learning for Medical Images [17.787804928943057]
  This paper presents a new way to identify additional positive pairs for BYOL, a state-of-the-art (SOTA) self-supervised learning framework. For each image, we select the most similar sample from other images as the additional positive and pull their features together with BYOL loss. Experimental results on two public medical datasets demonstrate that the proposed method can improve the classification performance.
  arXiv  Detail & Related papers  (2023-05-31T18:37:02Z)
• Contrastive Principal Component Learning: Modeling Similarity by Augmentation Overlap [50.48888534815361]
  We propose a novel Contrastive Principal Component Learning (CPCL) method composed of a contrastive-like loss and an on-the-fly projection loss. By CPCL, the learned low-dimensional embeddings theoretically preserve the similarity of augmentation distribution between samples.
  arXiv  Detail & Related papers  (2022-06-01T13:03:58Z)
• AdaPT-GMM: Powerful and robust covariate-assisted multiple testing [0.7614628596146599]
  We propose a new empirical Bayes method for co-assisted multiple testing with false discovery rate (FDR) control. Our method refines the adaptive p-value thresholding (AdaPT) procedure by generalizing its masking scheme. We show in extensive simulations and real data examples that our new method, which we call AdaPT-GMM, consistently delivers high power.
  arXiv  Detail & Related papers  (2021-06-30T05:06:18Z)
• Scalable Personalised Item Ranking through Parametric Density Estimation [53.44830012414444]
  Learning from implicit feedback is challenging because of the difficult nature of the one-class problem. Most conventional methods use a pairwise ranking approach and negative samplers to cope with the one-class problem. We propose a learning-to-rank approach, which achieves convergence speed comparable to the pointwise counterpart.
  arXiv  Detail & Related papers  (2021-05-11T03:38:16Z)
• Similarity Transfer for Knowledge Distillation [25.042405967561212]
  Knowledge distillation is a popular paradigm for learning portable neural networks by transferring the knowledge from a large model into a smaller one. We propose a novel method called similarity transfer for knowledge distillation (STKD), which aims to fully utilize the similarities between categories of multiple samples. It shows that STKD substantially has outperformed the vanilla knowledge distillation and has achieved superior accuracy over the state-of-the-art knowledge distillation methods.
  arXiv  Detail & Related papers  (2021-03-18T06:54:59Z)
• Doubly Contrastive Deep Clustering [135.7001508427597]
  We present a novel Doubly Contrastive Deep Clustering (DCDC) framework, which constructs contrastive loss over both sample and class views. Specifically, for the sample view, we set the class distribution of the original sample and its augmented version as positive sample pairs. For the class view, we build the positive and negative pairs from the sample distribution of the class. In this way, two contrastive losses successfully constrain the clustering results of mini-batch samples in both sample and class level.
  arXiv  Detail & Related papers  (2021-03-09T15:15:32Z)
• Almost-Matching-Exactly for Treatment Effect Estimation under Network Interference [73.23326654892963]
  We propose a matching method that recovers direct treatment effects from randomized experiments where units are connected in an observed network. Our method matches units almost exactly on counts of unique subgraphs within their neighborhood graphs.
  arXiv  Detail & Related papers  (2020-03-02T15:21:20Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.