The Visual Task Adaptation Benchmark

By Xiaohua Zhai, Joan Puigcerver, Alexander Kolesnikov, P. Ruyssen, Carlos Riquelme, Mario Lucic, Josip Djolonga, André Susano Pinto, Maxim Neumann, Alexey Dosovitskiy and 7 more. 2019

129 citations recorded by the source corpus.

Subfields: Evaluation & Benchmarks

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Evaluation & Benchmarks

Citations in the corpus

Self-Supervised Learning of Video-Induced Visual Invariances (2019, methodology): “We train the proposed model on the YouTube-8M (YT8M) data set (without us-ing video-level labels) and show that this approach leads to state-of-the-art self-supervised results on the 19 diverse downstream tasks of the Visual Task Adaptation Benchmark (VTAB) [65].”
Disentangling Transfer in Continual Reinforcement Learning (2022, background): “Interestingly, recent research [29, 49] suggests that there are still some gaps in our understanding of transfer learning.”
Scalable Transfer Learning with Expert Models (2020, methodology): “In VTAB-1k we use the recommended hyperparameter sweep and 800-training/200-validation split in [71].”
In-domain representation learning for remote sensing (2019, background): “These hyperparameter settings follow approximately the setup in (Zhai et al., 2019b) with minor modiﬁcations for the schedule and preprocessing.”
Large Scale Learning of General Visual Representations for Transfer (2019, methodology): “To further assess the generality of representations learned by BiT models, we leverage the recently introduced Visual Task Adaptation Benchmark (VTAB) [55] that consists of 19 visual tasks.”
Towards Learning a Universal Non-Semantic Representation of Speech (2020, background): “This approach of learning a strong representation on a large benchmark and then fine-tuning it for the downstream task has been proven very effective in other domains [1], and in this paper we demonstrate the same is true in the speech domain.”
A Survey of Deep Learning for Scientific Discovery (2020, background): “Note however that the precise eﬀects of transfer learning are not yet fully understood, and an active research area [116, 184, 266, 159, 143, 181, 235] looks at investigating its exact properties.”
How Useful Is Self-Supervised Pretraining for Visual Tasks? (2020, background): “There have also been a number of recent undertakings to provide comprehensive evaluation of today’s selfsupervised methods [28, 15, 25, 43, 24, 10, 33].”
The AI gambit: leveraging artificial intelligence to combat climate change—opportunities, challenges, and recommendations (2021, background): “0 (2020), Russakovsky (2015) and VTAB (2020), have been heavily focused on driving accuracy improvements with little regard for improving on energy efficiency (Perrault et al. 2019; Reddi et al. 2020).”
The Platonic Representation Hypothesis (2024, unknown): “2 (left), we bin these models based on their average transfer performance on the VTAB dataset (Zhai et al., 2019), and then measure the average kernel alignment of the models within each bin.”
The Origins and Prevalence of Texture Bias in Convolutional Neural Networks (2019, methodology): “For rotation, exemplar, and supervised losses, we used ResNet-50 v2 models made available as part of the Visual Task Adaptation Benchmark [104, 95].”