Duality Diagram Similarity: a generic framework for initialization selection in task transfer learning

TL;DR

This paper introduces a duality diagram similarity (DDS) framework for efficiently predicting the best pre-trained model for transfer learning tasks, significantly improving selection accuracy and speed.

Contribution

The paper presents a novel DDS-based method for model initialization selection in transfer learning, outperforming existing methods in accuracy and computational efficiency.

Findings

DDS correlates highly (0.86) with actual transfer performance.

The method reduces transfer ranking computation to under 2 minutes.

It is robust across different tasks and datasets, including Pascal VOC and NYUv2.

Abstract

In this paper, we tackle an open research question in transfer learning, which is selecting a model initialization to achieve high performance on a new task, given several pre-trained models. We propose a new highly efficient and accurate approach based on duality diagram similarity (DDS) between deep neural networks (DNNs). DDS is a generic framework to represent and compare data of different feature dimensions. We validate our approach on the Taskonomy dataset by measuring the correspondence between actual transfer learning performance rankings on 17 taskonomy tasks and predicted rankings. Computing DDS based ranking for $17\times17$ transfers requires less than 2 minutes and shows a high correlation ($0.86$) with actual transfer learning rankings, outperforming state-of-the-art methods by a large margin ($10\%$) on the Taskonomy benchmark. We also demonstrate the robustness of our model selection approach to a new task, namely Pascal VOC semantic segmentation. Additionally, we show that our method can be applied to select the best layer locations within a DNN for transfer learning on 2D, 3D and semantic tasks on NYUv2 and Pascal VOC datasets.