Picture for Yuhai Tu

Yuhai Tu

Maximal Domain Independent Representations Improve Transfer Learning

Add code
Jun 01, 2023
Figure 1 for Maximal Domain Independent Representations Improve Transfer Learning
Figure 2 for Maximal Domain Independent Representations Improve Transfer Learning
Figure 3 for Maximal Domain Independent Representations Improve Transfer Learning
Figure 4 for Maximal Domain Independent Representations Improve Transfer Learning
Viaarxiv icon

Effective Dynamics of Generative Adversarial Networks

Add code
Dec 08, 2022
Viaarxiv icon

Stochastic gradient descent introduces an effective landscape-dependent regularization favoring flat solutions

Add code
Jun 02, 2022
Figure 1 for Stochastic gradient descent introduces an effective landscape-dependent regularization favoring flat solutions
Figure 2 for Stochastic gradient descent introduces an effective landscape-dependent regularization favoring flat solutions
Figure 3 for Stochastic gradient descent introduces an effective landscape-dependent regularization favoring flat solutions
Viaarxiv icon

The activity-weight duality in feed forward neural networks: The geometric determinants of generalization

Add code
Mar 22, 2022
Figure 1 for The activity-weight duality in feed forward neural networks: The geometric determinants of generalization
Figure 2 for The activity-weight duality in feed forward neural networks: The geometric determinants of generalization
Figure 3 for The activity-weight duality in feed forward neural networks: The geometric determinants of generalization
Figure 4 for The activity-weight duality in feed forward neural networks: The geometric determinants of generalization
Viaarxiv icon

Loss Landscape Dependent Self-Adjusting Learning Rates in Decentralized Stochastic Gradient Descent

Add code
Dec 02, 2021
Figure 1 for Loss Landscape Dependent Self-Adjusting Learning Rates in Decentralized Stochastic Gradient Descent
Figure 2 for Loss Landscape Dependent Self-Adjusting Learning Rates in Decentralized Stochastic Gradient Descent
Figure 3 for Loss Landscape Dependent Self-Adjusting Learning Rates in Decentralized Stochastic Gradient Descent
Figure 4 for Loss Landscape Dependent Self-Adjusting Learning Rates in Decentralized Stochastic Gradient Descent
Viaarxiv icon

Phases of learning dynamics in artificial neural networks: with or without mislabeled data

Add code
Jan 16, 2021
Figure 1 for Phases of learning dynamics in artificial neural networks: with or without mislabeled data
Figure 2 for Phases of learning dynamics in artificial neural networks: with or without mislabeled data
Figure 3 for Phases of learning dynamics in artificial neural networks: with or without mislabeled data
Figure 4 for Phases of learning dynamics in artificial neural networks: with or without mislabeled data
Viaarxiv icon

How neural networks find generalizable solutions: Self-tuned annealing in deep learning

Add code
Jan 06, 2020
Figure 1 for How neural networks find generalizable solutions: Self-tuned annealing in deep learning
Figure 2 for How neural networks find generalizable solutions: Self-tuned annealing in deep learning
Figure 3 for How neural networks find generalizable solutions: Self-tuned annealing in deep learning
Figure 4 for How neural networks find generalizable solutions: Self-tuned annealing in deep learning
Viaarxiv icon

Continual Learning with Self-Organizing Maps

Add code
Apr 19, 2019
Figure 1 for Continual Learning with Self-Organizing Maps
Figure 2 for Continual Learning with Self-Organizing Maps
Figure 3 for Continual Learning with Self-Organizing Maps
Figure 4 for Continual Learning with Self-Organizing Maps
Viaarxiv icon

Learning to Learn without Forgetting By Maximizing Transfer and Minimizing Interference

Add code
Oct 29, 2018
Figure 1 for Learning to Learn without Forgetting By Maximizing Transfer and Minimizing Interference
Figure 2 for Learning to Learn without Forgetting By Maximizing Transfer and Minimizing Interference
Figure 3 for Learning to Learn without Forgetting By Maximizing Transfer and Minimizing Interference
Figure 4 for Learning to Learn without Forgetting By Maximizing Transfer and Minimizing Interference
Viaarxiv icon