Weakly-Supervised Disentanglement Without Compromises
- URL: http://arxiv.org/abs/2002.02886v4
- Date: Tue, 20 Oct 2020 15:22:16 GMT
- Title: Weakly-Supervised Disentanglement Without Compromises
- Authors: Francesco Locatello, Ben Poole, Gunnar R\"atsch, Bernhard Sch\"olkopf,
Olivier Bachem, Michael Tschannen
- Abstract summary: Intelligent agents should be able to learn useful representations by observing changes in their environment.
We model such observations as pairs of non-i.i.d. images sharing at least one of the underlying factors of variation.
We show that only knowing how many factors have changed, but not which ones, is sufficient to learn disentangled representations.
- Score: 53.55580957483103
- License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/
- Abstract: Intelligent agents should be able to learn useful representations by
observing changes in their environment. We model such observations as pairs of
non-i.i.d. images sharing at least one of the underlying factors of variation.
First, we theoretically show that only knowing how many factors have changed,
but not which ones, is sufficient to learn disentangled representations.
Second, we provide practical algorithms that learn disentangled representations
from pairs of images without requiring annotation of groups, individual
factors, or the number of factors that have changed. Third, we perform a
large-scale empirical study and show that such pairs of observations are
sufficient to reliably learn disentangled representations on several benchmark
data sets. Finally, we evaluate our learned representations and find that they
are simultaneously useful on a diverse suite of tasks, including generalization
under covariate shifts, fairness, and abstract reasoning. Overall, our results
demonstrate that weak supervision enables learning of useful disentangled
representations in realistic scenarios.
Related papers
- Multi-View Causal Representation Learning with Partial Observability [36.37049791756438]
We present a unified framework for studying identifiability of representations learned from simultaneously observed views.
We prove that the information shared across all subsets of any number of views can be learned up to a smooth bijection using contrastive learning.
We experimentally validate our claims on numerical, image, and multi-modal data sets.
arXiv Detail & Related papers (2023-11-07T15:07:08Z) - Leveraging sparse and shared feature activations for disentangled
representation learning [112.22699167017471]
We propose to leverage knowledge extracted from a diversified set of supervised tasks to learn a common disentangled representation.
We validate our approach on six real world distribution shift benchmarks, and different data modalities.
arXiv Detail & Related papers (2023-04-17T01:33:24Z) - On the Importance of Contrastive Loss in Multimodal Learning [34.91089650516183]
We analyze the training dynamics of a simple multimodal contrastive learning model.
We show that contrastive pairs are important for the model to efficiently balance the learned representations.
arXiv Detail & Related papers (2023-04-07T16:25:18Z) - An Empirical Investigation of Representation Learning for Imitation [76.48784376425911]
Recent work in vision, reinforcement learning, and NLP has shown that auxiliary representation learning objectives can reduce the need for large amounts of expensive, task-specific data.
We propose a modular framework for constructing representation learning algorithms, then use our framework to evaluate the utility of representation learning for imitation.
arXiv Detail & Related papers (2022-05-16T11:23:42Z) - RELAX: Representation Learning Explainability [10.831313203043514]
We propose RELAX, which is the first approach for attribution-based explanations of representations.
ReLAX explains representations by measuring similarities in the representation space between an input and masked out versions of itself.
We provide theoretical interpretations of RELAX and conduct a novel analysis of feature extractors trained using supervised and unsupervised learning.
arXiv Detail & Related papers (2021-12-19T14:51:31Z) - Desiderata for Representation Learning: A Causal Perspective [104.3711759578494]
We take a causal perspective on representation learning, formalizing non-spuriousness and efficiency (in supervised representation learning) and disentanglement (in unsupervised representation learning)
This yields computable metrics that can be used to assess the degree to which representations satisfy the desiderata of interest and learn non-spurious and disentangled representations from single observational datasets.
arXiv Detail & Related papers (2021-09-08T17:33:54Z) - Investigating the Role of Negatives in Contrastive Representation
Learning [59.30700308648194]
Noise contrastive learning is a popular technique for unsupervised representation learning.
We focus on disambiguating the role of one of these parameters: the number of negative examples.
We find that the results broadly agree with our theory, while our vision experiments are murkier with performance sometimes even being insensitive to the number of negatives.
arXiv Detail & Related papers (2021-06-18T06:44:16Z) - Odd-One-Out Representation Learning [1.6822770693792826]
We show that a weakly-supervised downstream task based on odd-one-out observations is suitable for model selection.
We also show that a bespoke metric-learning VAE model which performs highly on this task also out-performs other standard unsupervised and a weakly-supervised disentanglement model.
arXiv Detail & Related papers (2020-12-14T22:01:15Z) - A Sober Look at the Unsupervised Learning of Disentangled
Representations and their Evaluation [63.042651834453544]
We show that the unsupervised learning of disentangled representations is impossible without inductive biases on both the models and the data.
We observe that while the different methods successfully enforce properties "encouraged" by the corresponding losses, well-disentangled models seemingly cannot be identified without supervision.
Our results suggest that future work on disentanglement learning should be explicit about the role of inductive biases and (implicit) supervision.
arXiv Detail & Related papers (2020-10-27T10:17:15Z)
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.