Scalable Greedy Algorithms for Transfer Learning

Abstract

In this paper we consider the binary transfer learning problem, focusing on how to select and combine sources from a large pool to yield a good performance on a target task. Constraining our scenario to real world, we do not assume the direct access to the source data, but rather we employ the source hypotheses trained from them. We propose an efficient algorithm that selects relevant source hypotheses and feature dimensions simultaneously, building on the literature on the best subset selection problem. Our algorithm achieves state-of-the-art results on three computer vision datasets, substantially outperforming both transfer learning and popular feature selection baselines in a small-sample setting. We also present a randomized variant that achieves the same results with the computational cost independent from the number of source hypotheses and feature dimensions. Also, we theoretically prove that, under reasonable assumptions on the source hypotheses, our algorithm can learn effectively from few examples.

References in corpus (8)

• Caffe: Convolutional Architecture for Fast Feature Embedding
• DeCAF: A Deep Convolutional Activation Feature for Generic Visual Recognition
• Learning Transferable Features with Deep Adaptation Networks
• Unsupervised Domain Adaptation by Backpropagation
• Proceedings of the 29th International Conference on Machine Learning (ICML-12)
• Frustratingly Easy Domain Adaptation
• Submodular meets Spectral: Greedy Algorithms for Subset Selection, Sparse Approximation and Dictionary Selection
• When Naïve Bayes Nearest Neighbours Meet Convolutional Neural Networks

Cited by in corpus (3)

• Energy efficient distributed analytics at the edge of the network for IoT environments
• Boosting Binary Masks for Multi-Domain Learning through Affine Transformations
• Long-Term Ensemble Learning of Visual Place Classifiers