A Regularization Approach to Learning Task Relationships in Multitask Learning

2014 ◽  
Vol 8 (3) ◽  
pp. 1-31 ◽  
Author(s):  
Yu Zhang ◽  
Dit-Yan Yeung
Keyword(s):  
Learning Task ◽  
Multitask Learning ◽  
Approach To Learning

scholarly journals A Principled Approach for Learning Task Similarity in Multitask Learning

2019 ◽  
Author(s):  
Changjian Shui ◽  
Mahdieh Abbasi ◽  
Louis-Émile Robitaille ◽  
Boyu Wang ◽  
Christian Gagné
Keyword(s):  
Neural Network ◽  
Algorithm Design ◽  
Learning Task ◽  
Wasserstein Distance ◽  
Multitask Learning ◽  
Point Of View ◽  
Shared Knowledge ◽  
Theoretical Point ◽  
Task Similarity ◽  
Similarity Information

Multitask learning aims at solving a set of related tasks simultaneously, by exploiting the shared knowledge for improving the performance on individual tasks. Hence, an important aspect of multitask learning is to understand the similarities within a set of tasks. Previous works have incorporated this similarity information explicitly (e.g., weighted loss for each task) or implicitly (e.g., adversarial loss for feature adaptation), for achieving good empirical performances. However, the theoretical motivations for adding task similarity knowledge are often missing or incomplete. In this paper, we give a different perspective from a theoretical point of view to understand this practice. We first provide an upper bound on the generalization error of multitask learning, showing the benefit of explicit and implicit task similarity knowledge. We systematically derive the bounds based on two distinct task similarity metrics: H divergence and Wasserstein distance. From these theoretical results, we revisit the Adversarial Multi-task Neural Network, proposing a new training algorithm to learn the task relation coefficients and neural network parameters iteratively. We assess our new algorithm empirically on several benchmarks, showing not only that we find interesting and robust task relations, but that the proposed approach outperforms the baselines, reaffirming the benefits of theoretical insight in algorithm design.

scholarly journals Learning Task Sampling Policy for Multitask Learning

2021 ◽  
Author(s):  
Dhanasekar Sundararaman ◽  
Henry Tsai ◽  
Kuang-Huei Lee ◽  
Iulia Turc ◽  
Lawrence Carin
Keyword(s):  
Learning Task ◽  
Multitask Learning

Epistemic and relational conflicts in sharing identical vs. complementary information during cooperative learning

2002 ◽  
Vol 61 (3) ◽  
pp. 139-151 ◽  
Author(s):  
Céline Darnon ◽  
Céline Buchs ◽  
Fabrizio Butera
Keyword(s):  
Cooperative Learning ◽  
Social Comparison ◽  
Positive Relationship ◽  
Learning Task ◽  
Control Group ◽  
Complementary Information ◽  
Relational Conflict ◽  
The Social ◽  
The University

When interacting on a learning task, which is typical of several academic situations, individuals may experience two different motives: Understanding the problem, or showing their competences. When a conflict (confrontation of divergent propositions) emerges from this interaction, it can be solved either in an epistemic way (focused on the task) or in a relational way (focused on the social comparison of competences). The latter is believed to be detrimental for learning. Moreover, research on cooperative learning shows that when they share identical information, partners are led to compare to each other, and are less encouraged to cooperate than when they share complementary information. An epistemic vs. relational conflict vs. no conflict was provoked in dyads composed by a participant and a confederate, working either on identical or on complementary information (N = 122). Results showed that, if relational and epistemic conflicts both entailed more perceived interactions and divergence than the control group, only relational conflict entailed more perceived comparison activities and a less positive relationship than the control group. Epistemic conflict resulted in a more positive perceived relationship than the control group. As far as performance is concerned, relational conflict led to a worse learning than epistemic conflict, and - after a delay - than the control group. An interaction between the two variables on delayed performance showed that epistemic and relational conflicts were different only when working with complementary information. This study shows the importance of the quality of relationship when sharing information during cooperative learning, a crucial factor to be taken into account when planning educational settings at the university.

Blocking Is Sensitive to Causal Structure in 4-Year-Old and 8-Year-Old Children

2005 ◽  
Vol 52 (4) ◽  
pp. 264-271 ◽  
Author(s):  
Tom Beckers ◽  
Uschi Van den Broeck ◽  
Marij Renne ◽  
Stefaan Vandorpe ◽  
Jan De Houwer ◽  
...  
Keyword(s):  
Young Children ◽  
Associative Learning ◽  
Learning Theory ◽  
Causal Structure ◽  
Age Groups ◽  
Learning Task ◽  
Contingency Learning

Abstract. In a contingency learning task, 4-year-old and 8-year-old children had to predict the outcome displayed on the back of a card on the basis of cues presented on the front. The task was embedded in either a causal or a merely predictive scenario. Within this task, either a forward blocking or a backward blocking procedure was implemented. Blocking occurred in the causal but not in the predictive scenario. Moreover, blocking was affected by the scenario to the same extent in both age groups. The pattern of results was similar for forward and backward blocking. These results suggest that even young children are sensitive to the causal structure of a contingency learning task and that the occurrence of blocking in such a task defies an explanation in terms of associative learning theory.

Good Things Come to Those Who Wait: Enhancement in a Rule-Based Category Learning Task After Social-Evaluative Stress

2014 ◽  
Author(s):  
Shawn Ell ◽  
Steve Hutchinson ◽  
Lauren Hawthorne ◽  
Lauren Szymula ◽  
Shannon K. McCoy
Keyword(s):  
Category Learning ◽  
Learning Task ◽  
Rule Based
Sign in / Sign up

Export Citation Format

Share Document