Akshit Achara
Doctoral Candidate in Fairness, Robustness, and Medical Imaging AI at King's College London
I am a doctoral candidate with the Motion Modelling & Analysis Group at King’s College London and part of the DRIVE-Health programme, supervised by Prof. Andrew King, Prof. Alexander Hammers, and Dr. Esther Puyol Antón.
My research focuses on trustworthy machine learning for medical imaging, especially fairness, robustness, shortcut learning, representation alignment, and interpretability. More broadly, I am interested in understanding when models rely on spurious or unstable signals and in developing methods to detect, explain, and mitigate those failures.
Previously, I was a Research Engineer at GE Research in Bangalore, where I worked on NLP methods for timeline-based report summarisation and other applied machine learning problems. Outside research, I enjoy cricket, football, music, and catching up with friends.
news
| Mar 2, 2026 | Our paper entitled “Multi-Way Representation Alignment” has been published at the ICLR 2026 Re-Align Workshop. |
|---|---|
| Feb 28, 2026 | Our paper entitled “Right Regions, Wrong Labels: Semantic Label Flips in Segmentation under Correlation Shift” has been published at Catch, Adapt, and Operate: Monitoring ML Models Under Drift (ICLR 2026 CAO Workshop). |
| Dec 21, 2025 | Our paper entitled “Localising Shortcut Learning in Pixel Space via Ordinal Scoring Correlations for Attribution Representations (OSCAR)” is available on arXiv. |
| Sep 19, 2025 | Our paper entitled “Invisible Attributes, Visible Biases: Exploring Demographic Shortcuts in MRI-Based Alzheimer’s Disease Classification” has been published in FAIMI 2025. |
| Jan 22, 2025 | Our paper entitled “Watching the AI Watchdogs: A Fairness and Robustness Analysis of AI Safety Moderation Classifiers” has been accepted at the NAACL 2025 Main Conference. I am grateful for the mentorship I received from Dr. Anshuman Chhabra. |
| Oct 1, 2024 | I have started my PhD with the Motion Modelling & Analysis Group group at King’s College London as a part of the DRIVE-Health programme. I will be exploring algorithmic fairness in medical imaging. |
| Jun 25, 2024 | Our paper entitled “Efficient Biomedical Entity Linking: Clinical Text Standardization with Low-Resource Techniques” has been accepted at the BioNLP Workshop @ ACL 2024. |
| Jan 11, 2024 | Our paper entitled “Revealing the Underlying Patterns: Investigating Dataset Similarity, Performance, and Generalization” has been published at Neurocomputing. |
| Dec 8, 2023 | Our paper entitled “TrueDeep: A systematic approach of crack detection with less data” has been published at Expert Systems with Applications. |
| Jun 30, 2023 | Our paper entitled “CoreDeep: Improving Crack Detection Algorithms Using Width Stochasticity” has been accepted at CVIP, 2023. |