Clustering Learning Tasks and the Selective Cross-Task Transfer of Knowledge

By Sebastian Thrun, Joseph O'Sullivan. 1998

Subfields: Representation & Self-Supervised Learning

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Representation & Self-Supervised Learning

Citations in the corpus

Growing a Brain: Fine-Tuning by Increasing Model Capacity (2017, background): “Different from the non-parametric shallow models ( e.g ., nearest neighbors) that increase capacity when mem-orizing new data [40, 42], our developmental network cumulatively grows its capacity from novel tasks .”
Expert Gate: Lifelong Learning with a Network of Experts (2016, background): “To determine which related tasks to utilize, [35] cluster the tasks based on the mutual information gain when using the information from one task while learning another.”
SpotTune: Transfer Learning Through Adaptive Fine-Tuning (2018, background): “In the multi-task setting, knowing which tasks or parameters are shareable is a longstanding challenge [19, 23, 45, 29].”
Exploring the Challenges towards Lifelong Fact Learning (2018, background): “Each of these subsets is referred to as a “task”, originating from robotics applications [27].”
Multi-Task Learning for Spatio-Temporal Event Forecasting (2015, unknown): “We detail the three models below.”
Clustered Multi-Task Learning Via Alternating Structure Optimization (2011, background): “In [14], the mutual relatedness of tasks was estimated and know ledge of one task could be transferred to other tasks in the same cluster.”
Modeling disease progression via multi-task learning (2013, unknown): “, 2006), or the tasks are clustered into groups (Bakker and Heskes, 2003; Jacob et al., 2009; Thrun and O'Sullivan, 1998; Zhou et al., 2011).”
Regularized multi--task learning (2004, background): “There have also been various attempts to theoretically study multi–task learning, see [4, 5, 6, 7, 8, 15, 23].”
Discovering Structure in Multiple Learning Tasks: The TC Algorithm (1996, background): “Not shown here are results obtained in task familyT1, in which case TC performs approximately as well as its non-selective counterpart (see [29]).”
Online Methods for Multi-Domain Learning and Adaptation (2008, background): “Some multi-task work has also considered the grouping of tasks similar to our learning of domain subgroups (Thrun and O’Sullivan, 1998; Bakker and Heskes, 2003).”
Multi-domain learning by confidence-weighted parameter combination (2010, methodology): “Our clustering, and those considered by others (Thrun and O’Sullivan 1998; Bakker and Heskes 2003) assume a clustering of domains.”
Continual Learning in Neural Networks (2019, background): “To determine which related tasks to utilize, [151] clusters the tasks based on the mutual information gain when using the information from one task while learning another.”