Search

Few-shot classification (FSC) entails learning novel classes given only a few examples per class after a pre-training (or meta-training) phase on a set of base classes. Recent works have shown that simply fine-tuning a pre-trained Vision Transformer (ViT) on new test classes is a strong approach for FSC. Fine-tuning ViTs, however, is expensive in time, compute and storage. This has motivated the design of parameter efficient fine-tuning (PEFT) methods which fine-tune only a fraction of the Transformer's parameters. While these methods have shown promise, inconsistencies in experimental conditions make it difficult to disentangle their advantage from other experimental factors including the feature extractor architecture, pre-trained initialization and fine-tuning algorithm, amongst others. In our paper, we conduct a large-scale, experimentally consistent, empirical analysis to study PEFTs for few-shot image classification. Through a battery of over 1.8k controlled experiments on large-scale few-shot benchmarks including Meta-Dataset (MD) and ORBIT, we uncover novel insights on PEFTs that cast light on their efficacy in fine-tuning ViTs for few-shot classification. Through our controlled empirical study, we have two main findings: (i) Fine-tuning just the LayerNorm parameters (which we call LN-Tune) during few-shot adaptation is an extremely strong baseline across ViTs pre-trained with both self-supervised and supervised objectives, (ii) For self-supervised ViTs, we find that simply learning a set of scaling parameters for each attention matrix (which we call AttnScale) along with a domain-residual adapter (DRA) module leads to state-of-the-art performance (while being $\sim\!$ 9$\times$ more parameter-efficient) on MD. Our extensive empirical findings set strong baselines and call for rethinking the current design of PEFT methods for FSC.

Image-text contrastive models such as CLIP are useful for a variety of downstream applications including zero-shot classification, image-text retrieval and transfer learning. However, these contrastively trained vision-language models often fail on compositional visio-linguistic tasks such as Winoground with performance equivalent to random chance. In our paper, we address this issue and propose a sample-efficient light-weight method called SDS-CLIP to improve the compositional visio-linguistic reasoning capabilities of CLIP. The core idea of our method is to use differentiable image parameterizations to fine-tune CLIP with a distillation objective from large text-to-image generative models such as Stable-Diffusion which are relatively good at visio-linguistic reasoning tasks. On the challenging Winoground compositional reasoning benchmark, our method improves the absolute visio-linguistic performance of different CLIP models by up to 7%, while on the ARO dataset, our method improves the visio-linguistic performance by upto 3%. As a byproduct of inducing visio-linguistic reasoning into CLIP, we also find that the zero-shot performance improves marginally on a variety of downstream datasets. Our method reinforces that carefully designed distillation objectives from generative models can be leveraged to extend existing contrastive image-text models with improved visio-linguistic reasoning capabilities.

Federated learning has been predominantly concerned with collaborative training of deep networks from scratch, and especially the many challenges that arise, such as communication cost, robustness to heterogeneous data, and support for diverse device capabilities. However, there is no unified framework that addresses all these problems together. This paper studies the challenges and opportunities of exploiting pre-trained Transformer models in FL. In particular, we propose to efficiently adapt such pre-trained models by injecting a novel attention-based adapter module at each transformer block that both modulates the forward pass and makes an early prediction. Training only the lightweight adapter by FL leads to fast and communication-efficient learning even in the presence of heterogeneous data and devices. Extensive experiments on standard FL benchmarks, including CIFAR-100, FEMNIST and SpeechCommandsv2 demonstrate that this simple framework provides fast and accurate FL while supporting heterogenous device capabilities, efficient personalization, and scalable-cost anytime inference., 14 pages, 3 figures

In this paper, we describe a graph-based algorithm that uses the features obtained by a self-supervised transformer to detect and segment salient objects in images and videos. With this approach, the image patches that compose an image or video are organised into a fully connected graph, where the edge between each pair of patches is labeled with a similarity score between patches using features learned by the transformer. Detection and segmentation of salient objects is then formulated as a graph-cut problem and solved using the classical Normalized Cut algorithm. Despite the simplicity of this approach, it achieves state-of-the-art results on several common image and video detection and segmentation tasks. For unsupervised object discovery, this approach outperforms the competing approaches by a margin of 6.1%, 5.7%, and 2.6%, respectively, when tested with the VOC07, VOC12, and COCO20K datasets. For the unsupervised saliency detection task in images, this method improves the score for Intersection over Union (IoU) by 4.4%, 5.6% and 5.2%. When tested with the ECSSD, DUTS, and DUT-OMRON datasets, respectively, compared to current state-of-the-art techniques. This method also achieves competitive results for unsupervised video object segmentation tasks with the DAVIS, SegTV2, and FBMS datasets., arXiv admin note: substantial text overlap with arXiv:2202.11539

Recent sharpness-aware minimisation (SAM) is known to find flat minima which is beneficial for better generalisation with improved robustness. SAM essentially modifies the loss function by reporting the maximum loss value within the small neighborhood around the current iterate. However, it uses the Euclidean ball to define the neighborhood, which can be inaccurate since loss functions for neural networks are typically defined over probability distributions (e.g., class predictive probabilities), rendering the parameter space non Euclidean. In this paper we consider the information geometry of the model parameter space when defining the neighborhood, namely replacing SAM’s Euclidean balls with ellipsoids induced by the Fisher information. Our approach, dubbed Fisher SAM, defines more accurate neighborhood structures that conform to the intrinsic metric of the underlying statistical manifold. For instance, SAM may probe the worst-case loss value at either a too nearby or inappropriately distant point due to the ignorance of the parameter space geometry, which is avoided by our Fisher SAM. Another recent Adaptive SAM approach stretches/shrinks the Euclidean ball in accordance with the scale of the parameter magnitudes. This might be dangerous, potentially destroying the neighborhood structure. We demonstrate improved performance of the proposed Fisher SAM on several benchmark datasets/tasks.

We study the highly practical but comparatively under-studied problem of latent-domain adaptation, where a source model should be adapted to a target dataset that contains a mixture of unlabelled domain-relevant and domain-irrelevant examples. Furthermore, motivated by the requirements for data privacy and the need for embedded and resource-constrained devices of all kinds to adapt to local data distributions, we focus on the setting of feed-forward source-free domain adaptation, where adaptation should not require access to the source dataset, and also be back propagation-free. Our solution is to meta-learn a network capable of embedding the mixed-relevance target dataset and dynamically adapting inference for target examples using cross-attention. The resulting framework leads to consistent improvement on strong ERM baselines. We also show that our framework sometimes even improves on the upper bound of domain-supervised adaptation, where only domain-relevant instances are provided for adaptation. This suggests that human annotated domain labels may not always be optimal, and raises the possibility of doing better through automated instance selection., Shorter version accepted at the First Workshop of Pre-training: Perspectives, Pitfalls, and Paths Forward at ICML 2022

Recent sharpness-aware minimisation (SAM) is known to find flat minima which is beneficial for better generalisation with improved robustness. SAM essentially modifies the loss function by reporting the maximum loss value within the small neighborhood around the current iterate. However, it uses the Euclidean ball to define the neighborhood, which can be inaccurate since loss functions for neural networks are typically defined over probability distributions (e.g., class predictive probabilities), rendering the parameter space non Euclidean. In this paper we consider the information geometry of the model parameter space when defining the neighborhood, namely replacing SAM's Euclidean balls with ellipsoids induced by the Fisher information. Our approach, dubbed Fisher SAM, defines more accurate neighborhood structures that conform to the intrinsic metric of the underlying statistical manifold. For instance, SAM may probe the worst-case loss value at either a too nearby or inappropriately distant point due to the ignorance of the parameter space geometry, which is avoided by our Fisher SAM. Another recent Adaptive SAM approach stretches/shrinks the Euclidean ball in accordance with the scale of the parameter magnitudes. This might be dangerous, potentially destroying the neighborhood structure. We demonstrate improved performance of the proposed Fisher SAM on several benchmark datasets/tasks.

We propose a meta-learning approach that learns from multiple tasks in a transductive setting, by leveraging the unlabeled query set in addition to the support set to generate a more powerful model for each task. To develop our framework, we revisit the empirical Bayes formulation for multi-task learning. The evidence lower bound of the marginal log-likelihood of empirical Bayes decomposes as a sum of local KL divergences between the variational posterior and the true posterior on the query set of each task. We derive a novel amortized variational inference that couples all the variational posteriors via a meta-model, which consists of a synthetic gradient network and an initialization network. Each variational posterior is derived from synthetic gradient descent to approximate the true posterior on the query set, although where we do not have access to the true gradient. Our results on the Mini-ImageNet and CIFAR-FS benchmarks for episodic few-shot classification outperform previous state-of-the-art methods. Besides, we conduct two zero-shot learning experiments to further explore the potential of the synthetic gradient., Comment: ICLR 2020