Arjun Pankajakshan

Arjun Pankajakshan is a PhD researcher at the Centre for Digital Music, Queen Mary University of London, supervised by Dr. Emmanouil Benetos, Dr. Dan Stowell, and Dr. Helen Bear. His research focuses on computational analysis of sound scenes and events, particularly classification, segmentation, and sequence modeling of polyphonic sound events using machine learning. Previously, he served as a project associate at the MANAS Lab, Indian Institute of Technology Mandi, working on multi-modal bird classification under Dr. Padmanabhan Rajan and Dr. Arnav Bhavsar.

Research Interests:

• Machine Learning applications in audio processing
• Sound event detection and activity modeling
• Deepfake audio detection systems
• Polyphonic sound scene analysis

His recent work emphasizes neural collapse-based techniques for audio deepfake detection and real-time systems, alongside multi-task learning approaches for sound classification tasks. He has contributed to advancements in adversarial attack analysis, bird part localization in images, and audio tagging using novel noise modeling layers.

Affiliations:

• Queen Mary University of London: PhD researcher (current)
• MANAS Lab, IIT Mandi: Project associate (prior)

Labs/Teams: Active in the Centre for Digital Music and collaborated with MANAS Lab on multi-modal projects.