Born Again Neural Networks

Tommaso Furlanello, Zachary C. Lipton, Michael Tschannen, +2

arXiv (Cornell University)

Recent Advances in Autoencoder-Based Representation Learning

Michael Tschannen, Olivier Bachem, Mario Lučić

arXiv (Cornell University)

Artificial IntelligenceLarge Ai ModelStructured PredictionEngineeringMachine Learning +11

Born Again Neural Networks

Tommaso Furlanello, Zachary C. Lipton, Michael Tschannen, +2

CaltechAUTHORS (California Institute of Technology)

Artificial IntelligenceEngineeringMachine LearningDistillation ObjectivesRecurrent Neural Network +14

Soft-to-Hard Vector Quantization for End-to-End Learning Compressible Representations

Eirikur Agustsson, Fabian Mentzer, Michael Tschannen, +4

arXiv (Cornell University)

Artificial IntelligenceConvolutional Neural NetworkEngineeringMachine LearningQuantization (Signal Processing) +11

On Mutual Information Maximization for Representation Learning

Michael Tschannen, Josip Djolonga, Paul K. Rubenstein, +2

arXiv (Cornell University)

Artificial IntelligenceEngineeringMachine LearningAutoencodersMutual Information Maximization +15

High-Fidelity Generative Image Compression

Fabian Mentzer, George Toderici, Michael Tschannen, +1

arXiv (Cornell University)

Discriminator ArchitecturesLossy CompressionImage AnalysisMachine LearningEngineering +11

Soft-to-hard vector quantization for end-to-end learning compressible representations

Eirikur Agustsson, Fabian Mentzer, Michael Tschannen, +4

Lirias (KU Leuven)

A Large-scale Study of Representation Learning with the Visual Task Adaptation Benchmark

Xiaohua Zhai, Joan Puigcerver, Alexander Kolesnikov, +14

arXiv (Cornell University)

Multimodal LlmFew-shot LearningMachine VisionImage AnalysisMachine Learning +13

Scaling Vision Transformers to 22 Billion Parameters

Mostafa Dehghani, Josip Djolonga, Basil Mustafa, +38

arXiv (Cornell University)

EngineeringMachine LearningVision TransformersVideo InterpretationStable Training +15

Towards Image Understanding from Deep Compression without Decoding

Robert Torfason, Fabian Mentzer, Eirikur Agustsson, +3

arXiv (Cornell University)

Convolutional Neural NetworkEngineeringMachine LearningAutoencodersTowards Image +17