Lune

AAAI2021Top-tier venue

Towards Reusable Network Components by Learning Compatible Representations

Michael Gygli, Jasper R. R. Uijlings, Vittorio Ferrari

Abstract

This paper proposes to make a first step towards compatible and hence reusable network components. Rather than training networks for different tasks independently, we adapt the training process to produce network components that are compatible across tasks. In particular, we split a network into two components, a features extractor and a target task head, and propose various approaches to accomplish compatibility between them. We systematically analyse these approaches on the task of image classification on standard datasets. We demonstrate that we can produce components which are directly compatible without any fine-tuning or compromising accuracy on the original tasks. Afterwards, we demonstrate the use of compatible components on three applications: Unsupervised domain adaptation, transferring classifiers across feature extractors with different architectures, and increasing the computational efficiency of transfer learning.

Ask Lune

Know this paper in minutes, not an afternoon.

Lune can read every page of it and the top-tier papers that cite it, and gives you its claims, its evidence and its limits, each one cited to the paper it comes from. No skimming, no AI slop.

3 top-tier papers have cited it since. Know what they found before your reviewers do.

Questions to start from

Lune can draw on

the full paper · the top-tier papers that cite it · related work

Get the answer

Free to start. No credit card required.

Lineage

  • Data-Efficient Image Recognition with Contrastive Predictive Coding cites S4L: Self-Supervised Semi-Supervised Learning
  • S4L: Self-Supervised Semi-Supervised Learning cites Data-Efficient Image Recognition with Contrastive Predictive Coding
  • Self-Supervised Learning of Video-Induced Visual Invariances cites Data-Efficient Image Recognition with Contrastive Predictive Coding
  • Functional Alignment Can Mislead: Examining Model Stitching cites Relative representations enable zero-shot latent space communication
  • Latent Space Translation via Semantic Alignment cites Relative representations enable zero-shot latent space communication
Builds onCited byLeft to right by year; the more cited, the higher.

Cited by top-tier papers3

  1. Latent Space Translation via Semantic AlignmentValentino Maiorca, Luca Moschella, Antonio Norelli, Marco Fumero et al.NeurIPS 2023 · 59 citations
  2. Relative representations enable zero-shot latent space communicationLuca Moschella, Valentino Maiorca, Marco Fumero, Antonio Norelli et al.ICLR 2023 · 7 citations
  3. Functional Alignment Can Mislead: Examining Model StitchingDamian Smith, Harvey Mannering, Antonia MarcuICML 2025

Builds on5

  1. Data-Efficient Image Recognition with Contrastive Predictive CodingOlivier J. HénaffICML 2020 · 1,553 citations
  2. S4L: Self-Supervised Semi-Supervised LearningLucas Beyer, Xiaohua Zhai, Avital Oliver, Alexander KolesnikovICCV 2019 · 854 citations
  3. Task2Vec: Task Embedding for Meta-LearningAlessandro Achille, Michael Lam, Rahul Tewari, Avinash Ravichandran et al.ICCV 2019 · 359 citations
  4. Drop to Adapt: Learning Discriminative Features for Unsupervised Domain AdaptationSeungmin Lee, Dongwan Kim, Namil Kim, Seong-Gyun JeongICCV 2019 · 194 citations
  5. Self-Supervised Learning of Video-Induced Visual InvariancesMichael Tschannen, Josip Djolonga, Marvin Ritter, Aravindh Mahendran et al.CVPR 2020

Related papers

  1. PAS: Estimating the target accuracy before domain adaptationRaphaella Diniz, Jackson de Faria, Martin EsterICLR 2026 · 1 citation
  2. DRANet: Disentangling Representation and Adaptation Networks for Unsupervised Cross-Domain AdaptationSeunghun Lee, Sunghyun Cho, Sunghoon ImCVPR 2021
  3. Cooperative and Adversarial Learning: Co-enhancing Discriminability and Transferability in Domain AdaptationHui Sun, Zheng Xie, Xin-Ye Li, Ming LiAAAI 2023 · 5 citations
  4. Beyond Separability: Analyzing the Linear Transferability of Contrastive Representations to Related SubpopulationsJeff Z. HaoChen, Colin Wei, Ananya Kumar, Tengyu MaNeurIPS 2022 · 52 citations
  5. Adaptive Adversarial Network for Source-free Domain AdaptationHaifeng Xia, Handong Zhao, Zhengming DingICCV 2021 · 243 citations
Dusk over the sea between two cliffs drawn in fine vertical lines