Share This Author
Personalized Federated Learning using Hypernetworks
- Aviv Shamsian, Aviv Navon, Ethan Fetaya, Gal Chechik
- Computer ScienceInternational Conference on Machine Learning
- 8 March 2021
Since hypernetworks share information across clients, it is shown that pFedHN can generalize better to new clients whose distributions differ from any client observed during training, and decouples the communication cost from the trainable model size.
Learning the Pareto Front with Hypernetworks
- Aviv Navon, Aviv Shamsian, Gal Chechik, Ethan Fetaya
- Computer ScienceInternational Conference on Learning…
- 8 October 2020
The problem of learning the entire Pareto front, with the capability of selecting a desired operating point on the front after training is tackled, and PFL opens the door to new applications where models are selected based on preferences that are only available at run time.
Personalized Federated Learning with Gaussian Processes
- Idan Achituve, Aviv Shamsian, Aviv Navon, Gal Chechik, Ethan Fetaya
- Computer ScienceNeural Information Processing Systems
- 29 June 2021
Federated learning aims to learn a global model that performs well on client devices with limited cross-client communication. Personalized federated learning (PFL) further extends this setup to…
Auxiliary Learning by Implicit Differentiation
A novel framework, AuxiLearn, is proposed that targets both challenges of designing useful auxiliary tasks and combining auxiliary tasks into a single coherent loss, based on implicit differentiation.
Multi-Task Learning as a Bargaining Game
A new MTL optimization procedure, Nash-MTL, is described, which achieves state-of-the-art results on multiple MTL benchmarks in various domains and derives theoretical guarantees for its convergence.
GP-Tree: A Gaussian Process Classifier for Few-Shot Incremental Learning
- Idan Achituve, Aviv Navon, Yochai Yemini, Gal Chechik, Ethan Fetaya
- Computer ScienceICML
- 15 February 2021
This work develops a tree-based hierarchical model in which each internal node of the tree fits a GP to the data using the Pòlya-Gamma augmentation scheme, and shows how the general GP approach achieves improved accuracy on standard incremental few-shot learning benchmarks.
Capturing between-tasks covariance and similarities using multivariate linear mixed models
This paper proposes a procedure for constructing an estimator of a multivariate regression coefficient matrix that directly models and captures the within-group similarities, by employing aMultivariate linear mixed model formulation, with a joint estimation of covariance matrices for coefficients and errors via penalized likelihood.
A Study on the Evaluation of Generative Models
This study studies the evaluation metrics of generative models by generating a high-quality synthetic dataset on which classical metrics for comparison and shows that while FID and IS do correlate to several f-divergences, their ranking of close models can vary considerably making them problematic when used for fain-grained comparison.