Transfer Learning

A team from the Massachusetts Institute of Technology (MIT) and Google's artificial intelligence (AI) arm has found a way to use visual transfer learning to help robots grasp and manipulate objects more accurately. "We investigate whether existing pre-trained deep learning visual feature representations can improve the efficiency of learning robotic manipulation tasks, like grasping objects," write Google's Yen-Chen Lin and Andy Zeng of the research. "By studying how we can intelligently transfer neural network weights between vision models and affordance-based manipulation models, we can evaluate how different visual feature representations benefit the exploration process and enable robots to quickly acquire manipulation skills using different grippers. "We initialized our affordance-based manipulation models with backbones based on the ResNet-50 architecture and pre-trained on different vision tasks, including a classification model from ImageNet and a segmentation model from COCO. With different initialisations, the robot was then tasked with learning to grasp a diverse set of objects through trial and error.

We propose a new principle for transfer learning, based on a straightforward intuition: if two domains are similar to each other, the model trained on one domain should also perform well on the other domain, and vice versa. To formalize this intuition, we define the performance gap as a measure of the discrepancy between the source and target domains. We derive generalization bounds for the instance weighting approach to transfer learning, showing that the performance gap can be viewed as an algorithm-dependent regularizer, which controls the model complexity. Our theoretical analysis provides new insight into transfer learning and motivates a set of general, principled rules for designing new instance weighting schemes for transfer learning. These rules lead to gapBoost, a novel and principled boosting approach for transfer learning.

In few-shot learning, a machine learning system is required to learn from a small set of labelled examples of a specific task, such that it can achieve strong generalization on new unlabelled examples of the same task. Given the limited availability of labelled examples in such tasks, we need to make use of all the information we can. For this reason we propose the use of transductive meta-learning for few shot settings to obtain state-of-the-art few-shot learning. Usually a model learns task-specific information from a small training-set (the \emph{support-set}) and subsequently produces predictions on a small unlabelled validation set (\emph{target-set}). The target-set contains additional task-specific information which is not utilized by existing few-shot learning methods.

We aim to understand the value of additional labeled or unlabeled target data in transfer learning, for any given amount of source data; this is motivated by practical questions around minimizing sampling costs, whereby, target data is usually harder or costlier to acquire than source data, but can yield better accuracy. To this aim, we establish the first minimax-rates in terms of both source and target sample sizes, and show that performance limits are captured by new notions of discrepancy between source and target, which we refer to as transfer exponents. Interestingly, we find that attaining minimax performance is akin to ignoring one of the source or target samples, provided distributional parameters were known a priori. Moreover, we show that practical decisions -- w.r.t. Papers published at the Neural Information Processing Systems Conference.

Protein modeling is an increasingly popular area of machine learning research. Semi-supervised learning has emerged as an important paradigm in protein modeling due to the high cost of acquiring supervised protein labels, but the current literature is fragmented when it comes to datasets and standardized evaluation techniques. To facilitate progress in this field, we introduce the Tasks Assessing Protein Embeddings (TAPE), a set of five biologically relevant semi-supervised learning tasks spread across different domains of protein biology. We curate tasks into specific training, validation, and test splits to ensure that each task tests biologically relevant generalization that transfers to real-life scenarios. We benchmark a range of approaches to semi-supervised protein representation learning, which span recent work as well as canonical sequence learning techniques.

Humans and animals show remarkable flexibility in adjusting their behaviour when their goals, or rewards in the environment change. While such flexibility is a hallmark of intelligent behaviour, these multi-task scenarios remain an important challenge for machine learning algorithms and neurobiological models alike. We investigated two approaches that could enable this flexibility: factorized representations, which abstract away general aspects of a task from those prone to change, and nonparametric, memory-based approaches, which can provide a principled way of using similarity to past experiences to guide current behaviour. In particular, we combine the successor representation (SR), that factors the value of actions into expected outcomes and corresponding rewards, with evaluating task similarity through clustering the space of rewards. The proposed algorithm inverts a generative model over tasks, and dynamically samples from a flexible number of distinct SR maps while accumulating evidence about the current task context through amortized inference.

Transfer learning from natural image datasets, particularly ImageNet, using standard large models and corresponding pretrained weights has become a de-facto method for deep learning applications to medical imaging. However, there are fundamental differences in data sizes, features and task specifications between natural image classification and the target medical tasks, and there is little understanding of the effects of transfer. In this paper, we explore properties of transfer learning for medical imaging. A performance evaluation on two large scale medical imaging tasks shows that surprisingly, transfer offers little benefit to performance, and simple, lightweight models can perform comparably to ImageNet architectures. Investigating the learned representations and features, we find that some of the differences from transfer learning are due to the over-parametrization of standard models rather than sophisticated feature reuse.

Before sufficient training data is available, fine-tuning neural networks pre-trained on large-scale datasets substantially outperforms training from random initialization. However, fine-tuning methods suffer from two dilemmas, catastrophic forgetting and negative transfer. While several methods with explicit attempts to overcome catastrophic forgetting have been proposed, negative transfer is rarely delved into. In this paper, we launch an in-depth empirical investigation into negative transfer in fine-tuning and find that, for the weight parameters and feature representations, transferability of their spectral components is diverse. For safe transfer learning, we present Batch Spectral Shrinkage (BSS), a novel regularization approach to penalizing smaller singular values so that untransferable spectral components are suppressed.

Highlights: In this post we are going to show how to build a computer vision model without building it from scratch. The idea behind transfer learning is that a neural network that has been trained on a large dataset can apply its knowledge to a dataset that it has never seen before. That is, why it's called a transfer learning; we transfer the learning of an existing model to a new dataset. Previously we have explored how to improve the models performance using a data augmentation. The question now is, "what if we don't have enough data to train our network from scratch?".

Machine learning's ability to perform intellectually demanding tasks across various fields, materials science included, has caused it to receive considerable attention. Many believe that it could be used to unlock major time and cost savings in the development of new materials. The growing demand for the use of machine learning to derive fast-to-evaluate surrogate models of material properties has prompted scientists at the National Institute for Materials Science in Tsukuba, Japan, to demonstrate that it could be the key driver of the "next frontier" of materials science in recently published research. To learn, machines rely on processing data using both supervised and unsupervised learning. With no data, however, there is nothing to learn from.