Roman Malashin

According to our database, Roman Malashin authored at least 4 papers between 2021 and 2025.

Collaborative distances:

• Dijkstra number of seven.
• Erdős number of six.

2025

Neural network task specialization via domain constraining.

Roman Malashin, Daniil Ilyukhin

CoRR, April, 2025

Hypernym Bias: Unraveling Deep Classifier Training Dynamics through the Lens of Class Hierarchy.

Roman Malashin, Valeria Yachnaya, Alexander Mullin

CoRR, February, 2025

Hypernym Bias: Unraveling Deep Classifier Training Dynamics through the Lens of Class Hierarchy.

Roman Malashin, Valeria Yachnaya, Alexandr V. Mullin

Proceedings of the International Conference on Artificial Intelligence and Statistics, 2025

2021

Sparsely Ensembled Convolutional Neural Network Classifiers via Reinforcement Learning.

Roman Olegovich Malashin

Proceedings of the ICMLT 2021: 6th International Conference on Machine Learning Technologies, Jeju Island, Republic of Korea, April 23, 2021