Task representations in neural networks trained to perform many cognitive tasks

By Guangyu Robert Yang, Madhura R. Joglekar, H. F. Song, W. T. Newsome, Xiao-Jing Wang. 2019

606 citations recorded by the source corpus.

How researchers see this paper

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

In Cognitive Neuroscience

Citations in the corpus

Natural continual learning: success is a journey, not (just) a destination (2021, methodology): “In this section, we consider a set of neuroscience inspired ‘stimulus-response’ (SR) tasks (Yang et al., 2019; details in Appendix B).”
Continual Learning in Recurrent Neural Networks with Hypernetworks (2020, methodology): “Variants of SI have been used for different sequential datasets, but have not been systematically compared against other established methods [21, 7, 22].”
Computation Through Neural Population Dynamics (2020, methodology): “Yang et al. (2019) trained RNNs to perform 20 tasks commonly studied in systems neuroscience.”
Compositional Generalization for Primitive Substitutions (2019, background): “Human-level compositional learning has been an important open challenge (Yang et al., 2019), although there is a long history of studying compositionality in neural networks.”
Biological underpinnings for lifelong learning machines (2022, neutral): “167”
Deep neuroethology of a virtual rodent (2019, background): “While some efforts have been made to analyze neural activity in reduced systems trained to solve multiple tasks (Song et al., 2017; Yang et al., 2019), those studies lacked the important element of motor control in a physical environment.”
Continual learning in recurrent neural networks (2020, methodology): “Variants of SI have been used for different sequential datasets, but have not been systematically compared against other established methods (Yang et al., 2019; Masse et al., 2018; Lee, 2017).”
Interpreting neural computations by examining intrinsic and embedding dimensionality of neural activity (2021, background): “However, how these explanatory variables might map onto intrinsic variables and determine their dimensionality are among the most important unresolved questions at the intersection of machine learning and neuroscience [48].”
Organizing recurrent network dynamics by task-computation to enable continual learning (2020, methodology): “We demonstrate our approach on networks trained to perform multiple tasks akin to those studied by neuroscientists in animal models [9].”
DREAM Architecture: a Developmental Approach to Open-Ended Learning in Robotics (2020, neutral): “One interesting question is then: are prefrontal cortex mechanisms for state/action representational redescription completely different from those used to change/coordinate task-sets? Interestingly, neural representations of different cognitive tasks can be learned sequentially in a continual learni”
Compositional Continual Language Learning (2020, neutral): “Though these methods enable generalization when the training and test sentences have small difference, it has been an open problem (Yang et al., 2019) to reach human-level compositionality generalization.”
Multiple task-demands flexibly optimize neural geometry in human ventral temporal cortex (2026, unknown): “Failures to optimize representational geometries across repetitions - manifesting as insufficient alignment of the two types of representations that must be combined in this task - lead to behavioral errors and reduced accuracy.”
Radical flexibility of neural representation in frontoparietal cortex and the challenge of linking it to behaviour (2024, unknown): “[71,72]) or deriving network models from brain data to specify representational pathways to behaviour [73].”