"False"
Umeå universitet
printicon
Huvudmenyn dold.
Personalbild Anindya Sundar Das

Anindya Sundar Das

Kontakt

E-post
anindya.sundar.das@umu.se
Telefon
073-676 42 47

Verksam som

Anknytning
Doktorand vid Institutionen för datavetenskap
Plats
MIT-huset, Umeå universitet, MIT.D.248 Umeå universitet, 901 87 Umeå SE
Unmasking backdoors: an explainable defense via gradient-attention anomaly scoring for pre-trained language models
International Conference on Learning Representations 2026, ICLR 2026 : 26515-26531
Das, Anindya Sundar; Chen, Kangjie; Bhuyan, Monowar
Adaptive deviation learning for visual anomaly detection with data contamination
2025 IEEE/CVF Winter Conference on Applications of Computer Vision (WACV): proceedings, IEEE 2025 : 8863-8872
Das, Anindya Sundar; Pang, Guansong; Bhuyan, Monowar
Few-shot anomaly detection in text with deviation learning
Neural Information Processing. ICONIP 2023, Singapore: Springer 2024 : 425-438
Das, Anindya Sundar; Ajay, Aravind; Saha, Sriparna; et al.
Robust visual anomaly detection and localization under data contamination via multi-cue fusion
Das, Anindya Sundar; Elmroth, Erik; Bhuyan, Monowar
DriftWatch: Inference-time defense against backdoored LLM generation via internal anomaly signals
Das, Anindya Sundar; Soumit, Saha; Sriparna, Saha; et al.
Visa publikationer i DiVA