Posts by Collection

portfolio

Portfolio item number 1

Short description of portfolio item number 1

Portfolio item number 2

Short description of portfolio item number 2

publications

Low rank label subspace transformation for multi-label learning with missing labels

Published in Information Sciences, 2022

This paper proposes a unified framework, LRMML, that recovers missing labels in multi-label datasets by jointly leveraging auxiliary labels, low-rank constraints, and inter-label subspace separation to capture both local and global label correlations.

Recommended citation: Kumar, Sanjay, and Reshma Rastogi. "Low rank label subspace transformation for multi-label learning with missing labels. "Information Sciences (2022).

Discriminatory Label-specific Weights for Multi-label Learning with Missing Labels

Published in Neural Processing Letters, 2022

This paper proposes CIMML, a class imbalance-aware multi-label learning method for missing labels that estimates label weights based on observed, absent, and unobserved frequencies, and incorporates these into a weighted loss guided by auxiliary label correlations for improved label completion and classification.

Recommended citation: Rastogi, Reshma, and Sanjay Kumar. "Discriminatory label-specific weights for multi-label learning with missing labels. "Neural Processing Letters (2023).

Multi-label learning with missing labels using sparse global structure for label-specific features

Published in Applied Intelligence, 2023

This paper proposes a multi-label learning method for missing labels that identifies label-specific features constrained by a sparse global structure, enabling effective learning in high-dimensional spaces while leveraging supplementary label correlations for improved label recovery.

Recommended citation: Kumar, Sanjay, Nadira Ahmadi, and Reshma Rastogi. "Multi-label learning with missing labels using sparse global structure for label-specific features. "Applied Intelligence (2023).

Auxiliary label embedding for multi-label learning with missing labels

Published in Computer Vision and Machine Intelligence: Proceedings of CVMI 2022, 2023

This paper proposes a correlation embedding-based approach for multi-label learning with incomplete labels that learns an auxiliary label matrix and preserves label correlation structure in model coefficients to recover missing labels and improve classification performance.

Recommended citation: Kumar, Sanjay, and Reshma Rastogi. "Auxiliary label embedding for multi-label learning with missing labels." In Computer Vision and Machine Intelligence: Proceedings of CVMI 2022

Addressing Multi-Label Learning with Missing Labels via Feature Relevance guided Scaled Model Coefficients

Published in International Conference on Pattern Recognition, 2024

This paper proposes a multi-label learning method for missing labels that employs modified ℓ₂-norm regularization to encode feature relevance with global sparsity, incorporates label correlation learning, and uses squared hinge loss for robust classification.

Recommended citation: Kumar, Sanjay, and Reshma Rastogi. "Addressing Multi-Label Learning with Missing Labels via Feature Relevance guided Scaled Model Coefficients." In International Conference on Pattern Recognition, Springer Nature Switzerlan, 2024

GBSVR: Granular Ball Support Vector Regression

Published in arXiv preprint, 2025

This paper proposes GBSVR, a granular ball-based support vector regression method that reduces computational complexity and improves robustness to outliers by approximating data with granular balls and introducing a discretization strategy for continuous features.

Recommended citation: Rastogi, Reshma, Ankush Bisht, Sanjay Kumar, and Suresh Chandra. "GBSVR: Granular Ball Support Vector Regression. "arXiv preprint arXiv (2025).

teaching

Teaching experience 1

Undergraduate course, University 1, Department, 2014

This is a description of a teaching experience. You can use markdown like any other post.

Teaching experience 2

Workshop, University 1, Department, 2015