A major assumption in many machine learning and data mining algorithms is that the training and future data must be in the same feature space and have the same distribution. However, in many real-world applications, this assumption may not hold. For example, we sometimes have a classification task in one domain of interest, but we only have sufficient training data in another domain of interest, where the latter data may be in a different feature space or follow a different data distribution. In such cases, knowledge transfer, if done successfully, would greatly improve the performance of learning by avoiding much expensive data labeling efforts. In recent years, transfer learning has emerged as a new learning framework to address this problem. This survey focuses on categorizing and reviewing the current progress on transfer learning for classification, regression and clustering problems. In this survey, we discuss the relationship between transfer learning and other related machine learning techniques such as domain adaptation, multitask learning and sample selection bias, as well as co-variate shift. We also explore some potential future issues in transfer learning research.

Abstract. Transfer learning is the improvement of learning in a new task through the transfer of knowledge from a related task that has already been learned. While most machine learning algorithms are designed to address single tasks, the development of algorithms that facilitate transfer learning is a topic of ongoing interest in the machine-learning community. This chapter provides an introduction to the goals, formulations, and challenges of transfer learning. It surveys current research in this area, giving an overview of the state of the art and outlining the open problems. The survey covers transfer in both inductive learning and reinforcement learning, and discusses the issues of negative transfer and task mapping in depth.

Abstract. Most existing transfer learning techniques are limited to problems of knowledge transfer across tasks sharing the same set of class labels. In this paper, however, we relax this constraint and propose a spectral-based solution that aims at unveiling the intrinsic structure of the data and generating a partition of the target data, by transferring the eigenspace that well separates the source data. Furthermore, a clusteringbased KL divergence is proposed to automatically adjust how much to transfer. We evaluate the proposed model on text and image datasets where class categories of the source and target data are explicitly different, e.g., 3-classes transfer to 2-classes, and show that the proposed approach improves other baselines by an average of 10 % in accuracy. The source code and datasets are available from the authors. 1

Multi-task learning is a learning paradigm which seeks to improve the generalization performance of a learning task with the help of some other related tasks. In this paper, we propose a regularization formulation for learning the relationships between tasks in multi-task learning. This formulation can be viewed as a novel generalization of the regularization framework for single-task learning. Besides modeling positive task correlation, our method, called multi-task relationship learning (MTRL), can also describe negative task correlation and identify outlier tasks based on the same underlying principle. Under this regularization framework, the objective function of MTRL is convex. For efficiency, we use an alternating method to learn the optimal model parameters for each task as well as the relationships between tasks. We study MTRL in the symmetric multi-task learning setting and then generalize it to the asymmetric setting as well. We also study the relationships between MTRL and some existing multi-task learning methods. Experiments conducted on a toy problem as well as several benchmark data sets demonstrate the effectiveness of MTRL. 1

Maclin, to my committee members, and to the Machine Learning Group at the University of Wisconsin-Madison.

Abstract—The success of transfer to improve learning on a target task is highly dependent on the selected source data. Instance-based transfer methods reuse data from the source tasks to augment the training data for the target task. If poorly chosen, this source data may inhibit learning, resulting in negative transfer. The current best performing algorithm for instance-based transfer, TrAdaBoost, performs poorly when given irrelevant source data. We present a novel set-based boosting technique for instancebased transfer. The proposed algorithm, TransferBoost, boosts both individual instances and collective sets of instances from each source task. In effect, TransferBoost boosts each source task, assigning higher weight to those source tasks which show positive transferability to the target task, and then adjusts the weights of the instances within each source task via AdaBoost. The results demonstrate that TransferBoost significantly improves transfer performance over existing instancebased algorithms when given a mix of relevant and irrelevant source data. Keywords- knowledge transfer; boosting; ensemble methods I.

We present an interactive learning method that enables a user to iteratively refine a regression model. The user examines the output of the model, visualized as the vertical axis of a 2D scatterplot, and provides corrections by repositioning individual data instances to the correct output level. Each repositioned data instance acts as a control point for altering the learned model, using the geometry underlying the data. We capture the underlying structure of the data as a manifold, on which we compute a set of basis functions as the foundation for learning. Our results show that manifold-based interactive learning improves performance monotonically with each correction, outperforming alternative approaches.

We address the problem of learning multiple Bayesian network structures for experimental data where the experimental conditions define relationships among datasets. A metric of the relatedness of datasets, or tasks, can be described which contains valuable information that we exploit to learn more robust structures for each task. We represent the task-relatedness with an undirected graph. Our method uses a regularization framework over this task-relatedness graph to learn Bayesian network structures for each task that are smoothed toward the structures of related tasks. Experiments on synthetic data and real fMRI experiment data show that this method learns structures which are close to ground truth, when available, and which generalize to holdout data better than an existing multitask learning method, learning networks independently, and learning one global network for all tasks. 1

We address the problem of learning Bayesian networks for a collection of unsupervised tasks when limited data is available and a metric of the relatedness of tasks is given. We exploit this valuable information about taskrelatedness to learn more robust structures for each task than those learned with a standard multitask learning algorithm. Our approach is the first network structure learning algorithm that addresses zero-data learning; where no training data is available for a task, but data for related tasks and information about how the tasks are related is given. This paper describes our Task Relationship Aware Multitask (TRAM) algorithm for learning Bayesian networks. The inputs to the algorithm are task-specific sets of data and an undirected graph representing a metric of task-relatedness. TRAM regularizes over the task-relatedness graph to output a network for each task, each of which is biased toward the structures of related tasks. We empirically compare, using synthetic and neuroimaging data, TRAM with three baseline methods and show that task-relatedness knowledge impacts all learning methods. Only TRAM effectively uses this knowledge to learn more robust structures.

A Bayesian network is a standard tool in statistical data mining that gives a compact representation of relationships among variables in data. The structure of the network, or edges in the graph, represent conditional dependencies between variables. Machine learning algorithms have been developed to find Bayesian networks that model the underlying structure of relationships among data variables. For example, the variables could be the activity in various regions of the brain. Modern neuro-imaging technology allows us to look for patterns among the relationships between the activity of brain regions. Neuro-scientists suspect that certain mental illnesses, such as schizophrenia, exhibit abnormal brain networks. To explore these differences in networks, a different Bayesian network should be learned for each sub-population of subjects. Learning several different, but related, networks is known as multitask learning. The data is partitioned into tasks, and for each task a Bayesian network is learned. Simply partitioning the data and learning completely independent networks is problematic when there is not enough data in each task. Furthermore, we believe that the data is actually generated by similar, rather than completely independent, networks. In fact, we often have domain information about how tasks are related. For example, Subject A has a family history of schizophrenia, is