Subscribe to the gold package and get unlimited access to Shamra Academy
Register a new userContrastive Learning Inverts the Data Generating Process
293
0
0.0
(
0
)
Added by
Roland Zimmermann
Publication date
2021
fields
Informatics Engineering
and research's language is
English
Ask ChatGPT about the research
No Arabic abstract
Contrastive learning has recently seen tremendous success in self-supervised learning. So far, however, it is largely unclear why the learned representations generalize so effectively to a large variety of downstream tasks. We here prove that feedforward models trained with objectives belonging to the commonly used InfoNCE family learn to implicitly invert the underlying generative model of the observed data. While the proofs make certain statistical assumptions about the generative model, we observe empirically that our findings hold even if these assumptions are severely violated. Our theory highlights a fundamental connection between contrastive learning, generative modeling, and nonlinear independent component analysis, thereby furthering our understanding of the learned representations as well as providing a theoretical foundation to derive more effective contrastive losses.
rate research
Read More
How can neural networks trained by contrastive learning extract features from the unlabeled data? Why does contrastive learning usually need much stronger data augmentations than supervised learning to ensure good representations? These questions involve both the optimization and statistical aspects of deep learning, but can hardly be answered by analyzing supervised learning, where the target functions are the highest pursuit. Indeed, in self-supervised learning, it is inevitable to relate to the optimization/generalization of neural networks to how they can encode the latent structures in the data, which we refer to as the feature learning process. In this work, we formally study how contrastive learning learns the feature representations for neural networks by analyzing its feature learning process. We consider the case where our data are comprised of two types of features: the more semantically aligned sparse features which we want to learn from, and the other dense features we want to avoid. Theoretically, we prove that contrastive learning using $mathbf{ReLU}$ networks provably learns the desired sparse features if proper augmentations are adopted. We present an underlying principle called $textbf{feature decoupling}$ to explain the effects of augmentations, where we theoretically characterize how augmentations can reduce the correlations of dense features between positive samples while keeping the correlations of sparse features intact, thereby forcing the neural networks to learn from the self-supervision of sparse features. Empirically, we verified that the feature decoupling principle matches the underlying mechanism of contrastive learning in practice.
Contrastive learning applied to self-supervised representation learning has seen a resurgence in recent years, leading to state of the art performance in the unsupervised training of deep image models. Modern batch contrastive approaches subsume or significantly outperform traditional contrastive losses such as triplet, max-margin and the N-pairs loss. In this work, we extend the self-supervised batch contrastive approach to the fully-supervised setting, allowing us to effectively leverage label information. Clusters of points belonging to the same class are pulled together in embedding space, while simultaneously pushing apart clusters of samples from different classes. We analyze two possib
With the advent of big data across multiple high-impact applications, we are often facing the challenge of complex heterogeneity. The newly collected data usually consist of multiple modalities and characterized with multiple labels, thus exhibiting the co-existence of multiple types of heterogeneity. Although state-of-the-art techniques are good at modeling the complex heterogeneity with sufficient label information, such label information can be quite expensive to obtain in real applications, leading to sub-optimal performance using these techniques. Inspired by the capability of contrastive learning to utilize rich unlabeled data for improving performance, in this paper, we propose a unified heterogeneous learning framework, which combines both weighted unsupervised contrastive loss and weighted supervised contrastive loss to model multiple types of heterogeneity. We also provide theoretical analyses showing that the proposed weighted supervised contrastive loss is the lower bound of the mutual information of two samples from the same class and the weighted unsupervised contrastive loss is the lower bound of the mutual information between the hidden representation of two views of the same sample. Experimental results on real-world data sets demonstrate the effectiveness and the efficiency of the proposed method modeling multiple types of heterogeneity.
Contrastive learning (CL) is effective in learning data representations without label supervision, where the encoder needs to contrast each positive sample over multiple negative samples via a one-vs-many softmax cross-entropy loss. However, conventional CL is sensitive to how many negative samples are included and how they are selected. Proposed in this paper is a doubly CL strategy that contrasts positive samples and negative ones within themselves separately. We realize this strategy with contrastive attraction and contrastive repulsion (CACR) makes the query not only exert a greater force to attract more distant positive samples but also do so to repel closer negative samples. Theoretical analysis reveals the connection between CACR and CL from the perspectives of both positive attraction and negative repulsion and shows the benefits in both efficiency and robustness brought by separately contrasting within the sampled positive and negative pairs. Extensive large-scale experiments on standard vision tasks show that CACR not only consistently outperforms existing CL methods on benchmark datasets in representation learning, but also provides interpretable contrastive weights, demonstrating the efficacy of the proposed doubly contrastive strategy.
Deep neural nets typically perform end-to-end backpropagation to learn the weights, a procedure that creates synchronization constraints in the weight update step across layers and is not biologically plausible. Recent advances in unsupervised contrastive representation learning point to the question of whether a learning algorithm can also be made local, that is, the updates of lower layers do not directly depend on the computation of upper layers. While Greedy InfoMax separately learns each block with a local objective, we found that it consistently hurts readout accuracy in state-of-the-art unsupervised contrastive learning algorithms, possibly due to the greedy objective as well as gradient isolation. In this work, we discover that by overlapping local blocks stacking on top of each other, we effectively increase the decoder depth and allow upper blocks to implicitly send feedbacks to lower blocks. This simple design closes the performance gap between local learning and end-to-end contrastive learning algorithms for the first time. Aside from standard ImageNet experiments, we also show results on complex downstream tasks such as object detection and instance segmentation directly using readout features.
suggested questions
Log in to be able to interact and post comments
comments
Fetching comments
Sign in to be able to follow your search criteria
