Alert button
Picture for Sushrut Thorat

Sushrut Thorat

Alert button

Balancing stability and plasticity in continual learning: the readout-decomposition of activation change (RDAC) framework

Oct 10, 2023
Daniel Anthes, Sushrut Thorat, Peter König, Tim C. Kietzmann

Figure 1 for Balancing stability and plasticity in continual learning: the readout-decomposition of activation change (RDAC) framework
Figure 2 for Balancing stability and plasticity in continual learning: the readout-decomposition of activation change (RDAC) framework
Figure 3 for Balancing stability and plasticity in continual learning: the readout-decomposition of activation change (RDAC) framework
Figure 4 for Balancing stability and plasticity in continual learning: the readout-decomposition of activation change (RDAC) framework
Viaarxiv icon

Diagnosing Catastrophe: Large parts of accuracy loss in continual learning can be accounted for by readout misalignment

Oct 09, 2023
Daniel Anthes, Sushrut Thorat, Peter König, Tim C. Kietzmann

Figure 1 for Diagnosing Catastrophe: Large parts of accuracy loss in continual learning can be accounted for by readout misalignment
Viaarxiv icon

Characterising representation dynamics in recurrent neural networks for object recognition

Aug 23, 2023
Sushrut Thorat, Adrien Doerig, Tim C. Kietzmann

Figure 1 for Characterising representation dynamics in recurrent neural networks for object recognition
Figure 2 for Characterising representation dynamics in recurrent neural networks for object recognition
Figure 3 for Characterising representation dynamics in recurrent neural networks for object recognition
Figure 4 for Characterising representation dynamics in recurrent neural networks for object recognition
Viaarxiv icon

Category-orthogonal object features guide information processing in recurrent neural networks trained for object categorization

Nov 15, 2021
Sushrut Thorat, Giacomo Aldegheri, Tim C. Kietzmann

Figure 1 for Category-orthogonal object features guide information processing in recurrent neural networks trained for object categorization
Figure 2 for Category-orthogonal object features guide information processing in recurrent neural networks trained for object categorization
Figure 3 for Category-orthogonal object features guide information processing in recurrent neural networks trained for object categorization
Figure 4 for Category-orthogonal object features guide information processing in recurrent neural networks trained for object categorization
Viaarxiv icon

Modulation of early visual processing alleviates capacity limits in solving multiple tasks

Jul 30, 2019
Sushrut Thorat, Giacomo Aldegheri, Marcel A. J. van Gerven, Marius V. Peelen

Figure 1 for Modulation of early visual processing alleviates capacity limits in solving multiple tasks
Figure 2 for Modulation of early visual processing alleviates capacity limits in solving multiple tasks
Viaarxiv icon

The functional role of cue-driven feature-based feedback in object recognition

Mar 25, 2019
Sushrut Thorat, Marcel van Gerven, Marius Peelen

Figure 1 for The functional role of cue-driven feature-based feedback in object recognition
Figure 2 for The functional role of cue-driven feature-based feedback in object recognition
Figure 3 for The functional role of cue-driven feature-based feedback in object recognition
Figure 4 for The functional role of cue-driven feature-based feedback in object recognition
Viaarxiv icon

Implementing a Reverse Dictionary, based on word definitions, using a Node-Graph Architecture

Dec 17, 2016
Sushrut Thorat, Varad Choudhari

Figure 1 for Implementing a Reverse Dictionary, based on word definitions, using a Node-Graph Architecture
Figure 2 for Implementing a Reverse Dictionary, based on word definitions, using a Node-Graph Architecture
Figure 3 for Implementing a Reverse Dictionary, based on word definitions, using a Node-Graph Architecture
Figure 4 for Implementing a Reverse Dictionary, based on word definitions, using a Node-Graph Architecture
Viaarxiv icon