Robik Shrestha
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Robik Shrestha has a total of 16 co-authors. Here are the top co-authors by number of shared publications:
- Christopher Kanan
University of Rochester
5 shared publications - Kushal Kafle
Adobe Research
4 shared publications A. Mukherji
1 shared publicationAjay Divakaran
SRI International
1 shared publication- Arijit Ray
Department of Computer Science Boston University
1 shared publication H. Kulkarni
1 shared publication
Latest Publications
- Open Access
OccamNets: Mitigating Dataset Bias by Favoring Simpler Hypotheses
Dataset bias and spurious correlations can significantly impair generalization in deep neural networks. Many prior efforts have addressed this problem using either alternative loss functions or sampling strategies that focus on rare patterns. We propose a new direction: modifying the network architecture to impose inductive biases that make the network robust to dataset bias. Specifically, we propose OccamNets, which are biased to favor simpler solutions by design. OccamNets have two inductive biases. First, they are biased to use as little network depth as needed for an individual example. Second, they are biased toward using fewer image locations for prediction. While OccamNets are biased toward simpler hypotheses, they can learn more complex hypotheses if necessary. In experiments, OccamNets outperform or rival state-of-the-art methods run on architectures that do not incorporate these inductive biases. Furthermore, we demonstrate that when the state-of-the-art debiasing methods are combined with OccamNets results further improve.
Open AccessOccamNets: Mitigating Dataset Bias by Favoring Simpler Hypotheses
Dataset bias and spurious correlations can significantly impair generalization in deep neural networks. Many prior efforts have addressed this problem using either alternative loss functions or sampling strategies that focus on rare patterns. We propose a new direction: modifying the network architecture to impose inductive biases that make the network robust to dataset bias. Specifically, we propose OccamNets, which are biased to favor simpler solutions by design. OccamNets have two inductive biases. First, they are biased to use as little network depth as needed for an individual example. Second, they are biased toward using fewer image locations for prediction. While OccamNets are biased toward simpler hypotheses, they can learn more complex hypotheses if necessary. In experiments, OccamNets outperform or rival state-of-the-art methods run on architectures that do not incorporate these inductive biases. Furthermore, we demonstrate that when the state-of-the-art debiasing methods are combined with OccamNets results further improve.
- Open Access
- 1 January 2024
- Published by Association for Computational Linguistics (ACL)
- p. 14905-14918
BloomVQA: Assessing Hierarchical Multi-modal Comprehension
Yunye Gong, Robik Shrestha, Jared Claypoole, Michael Cogswell, Arijit Ray, Christopher Kanan, Ajay Divakaran. Findings of the Association for Computational Linguistics ACL 2024. 2024.
- 1 January 2024
- Published by Association for Computational Linguistics (ACL)
- p. 14905-14918
Open AccessBloomVQA: Assessing Hierarchical Multi-modal Comprehension
Yunye Gong, Robik Shrestha, Jared Claypoole, Michael Cogswell, Arijit Ray, Christopher Kanan, Ajay Divakaran. Findings of the Association for Computational Linguistics ACL 2024. 2024.
- 20 October 2022
- Published by Springer Nature
- p. 702-721
OccamNets: Mitigating Dataset Bias by Favoring Simpler Hypotheses
- 20 October 2022
- Published by Springer Nature
- p. 702-721
OccamNets: Mitigating Dataset Bias by Favoring Simpler Hypotheses
- 7 November 2020
- Published by Springer Nature
- p. 466-483
REMIND Your Neural Network to Prevent Catastrophic Forgetting
- 7 November 2020
- Published by Springer Nature
- p. 466-483
REMIND Your Neural Network to Prevent Catastrophic Forgetting
- Open Access
Answer Them All! Toward Universal Visual Question Answering Models
Visual Question Answering (VQA) research is split into two camps: the first focuses on VQA datasets that require natural image understanding and the second focuses on synthetic datasets that test reasoning. A good VQA algorithm should be capable of both, but only a few VQA algorithms are tested in this manner. We compare five state-of-the-art VQA algorithms across eight VQA datasets covering both domains. To make the comparison fair, all of the models are standardized as much as possible, e.g., they use the same visual features, answer vocabularies, etc. We find that methods do not generalize across the two domains. To address this problem, we propose a new VQA algorithm that rivals or exceeds the state-of-the-art for both domains.
Open AccessAnswer Them All! Toward Universal Visual Question Answering Models
Visual Question Answering (VQA) research is split into two camps: the first focuses on VQA datasets that require natural image understanding and the second focuses on synthetic datasets that test reasoning. A good VQA algorithm should be capable of both, but only a few VQA algorithms are tested in this manner. We compare five state-of-the-art VQA algorithms across eight VQA datasets covering both domains. To make the comparison fair, all of the models are standardized as much as possible, e.g., they use the same visual features, answer vocabularies, etc. We find that methods do not generalize across the two domains. To address this problem, we propose a new VQA algorithm that rivals or exceeds the state-of-the-art for both domains.
- 1 January 2018
- Published by International Centre for Integrated Mountain Development (ICIMOD)
Protocol for reviving Springs in the Hindu Kush Himalayas: A Practitioner's Manual
This handbook takes on the earlier attempts of RM&DD, and modifies it to suit the generic requirement of all ICIMOD RMCs. ICIMOD and ACWADAM, following a consultative process with major partners came up with detailed steps, which were then vetted at a workshop held in Gangtok, Sikkim, India in November 2015. This protocol is useful and practical, because: It incorporates both ‘hydrogeology’ and ‘socio-economic and governance’ issues to come up with a comprehensive understanding of springs and springsheds It combines aspects of research and knowledge generation (Steps 1 to 4) and implementation (steps 5 and 6). For those, who are interested only in knowledge generation can adopt the first four steps, but for implementation, all six steps are needed. It is a relatively easy to do, step by step approach that can be adopted by a diverse range of stakeholders – field implementers, grass root workers and NGOs and researchers. This manual provides a step by step approach, which taken together with a 2 week long practical classroom and field based training, will equip field level officials to implement spring revival programmes in their respective areas. This manual provide conceptual clarity around issues of spring management and revival.
- 1 January 2018
- Published by International Centre for Integrated Mountain Development (ICIMOD)
Protocol for reviving Springs in the Hindu Kush Himalayas: A Practitioner's Manual
This handbook takes on the earlier attempts of RM&DD, and modifies it to suit the generic requirement of all ICIMOD RMCs. ICIMOD and ACWADAM, following a consultative process with major partners came up with detailed steps, which were then vetted at a workshop held in Gangtok, Sikkim, India in November 2015. This protocol is useful and practical, because: It incorporates both ‘hydrogeology’ and ‘socio-economic and governance’ issues to come up with a comprehensive understanding of springs and springsheds It combines aspects of research and knowledge generation (Steps 1 to 4) and implementation (steps 5 and 6). For those, who are interested only in knowledge generation can adopt the first four steps, but for implementation, all six steps are needed. It is a relatively easy to do, step by step approach that can be adopted by a diverse range of stakeholders – field implementers, grass root workers and NGOs and researchers. This manual provides a step by step approach, which taken together with a 2 week long practical classroom and field based training, will equip field level officials to implement spring revival programmes in their respective areas. This manual provide conceptual clarity around issues of spring management and revival.