Publications

Recent preprints / Papers in submission

1. M. Zarvandi, M. Timothy, T. Wasserer, D. Ghoshdastidar. Interpretable kernel representation learning at scale: A unified framework utilizing Nyström approximation [Preprint]
2. N. Ayday, M. Sabanayagam, D. Ghoshdastidar. Why does your graph neural network fail on some graphs? Insights from exact generalisation error. [Preprint]
3. J. Ham, M. Fleissner, D. Ghoshdastidar. Impact of bottleneck layers and skip connections on the generalization of linear denoising autoencoders. [Preprint]
4. Y. Han, D. Ghoshdastidar. Attention learning is needed to efficiently learn parity function [Preprint]
5. M. Sabanayagam, R. Tsuchida, C. S. Ong, D. Ghoshdastidar. Generalization certificates for adversarially robust Bayesian linear regression. [Preprint]
6. S. Libo Feigin, M. Fleissner, D. Ghoshdastidar. Data augmentations go beyond encoding invariances: A theoretical study on self-supervised learning. [Preprint]
7. M. Fleissner, M. Zarvandi, D. Ghoshdastidar. Decision trees for interpretable clusters in mixture models and deep representations. [Preprint]
8. A. Craciun, D. Ghoshdastidar. On the stability of gradient descent for large learning rate. [Preprint]

Book chapters / Perspective articles

1. P. Esser, M. Fleissner, D. Ghoshdastidar. Theoretical foundations of representation learning using unlabeled data: Statistics and optimization. [Preprint] 

Contributing articles (peer reviewed)

1. T. Wasserer, M. Fleissner, D. Ghoshdastidar. PAC-Bayesian Analysis of the Surrogate Relation between Joint Embedding and Supervised Downstream Losses. ALT 2026. [Paper]
2. S. Mukherjee, S. S. Mukherjee, D. Ghoshdastidar. Wasserstein projection pursuit of non-Gaussian signals. Journal of Multivariate Analysis (special issue on Dimension Reduction in Multivariate Analysis), Article 105535, 2026 [Paper]
3. M. Eppert, S. Mukherjee, D. Ghoshdastidar. Recovering imbalanced clusters via gradient based projection pursuit. Journal of Multivariate Analysis (special issue on Dimension Reduction in Multivariate Analysis), Article 105530, 2026 [Paper] [Preprint]
4. A. van Elst, D. Ghoshdastidar. Tight PAC-Bayesian risk certificates for contrastive learning. SIAM Journal on Data Science, 7(4):1904-1927, 2025. [Paper] [Preprint]
5. A. Craciun, D. Ghoshdastidar. Non-Singularity of the Gradient Descent Map for Neural Networks with Piecewise Analytic Activations. NeurIPS 2025. [Paper]
6. L. Gosch, M. Sabanayagam, D. Ghoshdastidar, S. Günnemann. Provable robustness of (graph) neural networks against data poisoning and backdoor attacks. Transactions of Machine Learning, 2025. [Paper] [Preprint] (preliminary version received best paper award in AdvML-Frontiers @NeurIPS-2024)
7. M. Sabanayagam, L. Gosch, S. Günnemann, D. Ghoshdastidar. Exact certification of (graph) neural networks against label poisoning. ICLR 2025. [Paper] [Preprint] (spotlight, top 5%)
8. M. Fleissner, G. G. Anil, D. Ghoshdastidar. Infinite width limits of self supervised neural networks. AISTATS 2025. [Paper] [Preprint]
9. G. G. Anil, P. Esser, D. Ghoshdastidar. When can we approximate wide contrastive models with neural tangent kernels and principal component analysis? AAAI 2025. [Preprint]
10. L. Rendsburg, L. C. Vankadara, U. von Luxburg, D. Ghoshdastidar. A consistent estimator for confounding strength. Mathematical Statistics and Learning, 7 (3/4): 189-220, 2024. [Paper] [Preprint]
11. A. Singh, M. Sabanayagam, K. Muandet, D. Ghoshdastidar. Fast adaptive test-time defense with robust features. Transactions of Machine Learning, 2024. [Paper] [Preprint]
12. P. Esser, S. Mukherjee, D. Ghoshdastidar. Representation learning dynamics of self-supervised models. Transactions of Machine Learning, 2024. [Paper] [Preprint]
13. M. Fleissner, L. Vankadara, D. Ghoshdastidar. Explaining kernel clustering via decision trees. ICLR 2024 [Paper]
14. P. Esser, M. Fleissner, D. Ghoshdastidar. Non-parametric representation learning with kernels. AAAI 2024. [Paper] [Preprint]
15. M. Sabanayagam, P. Esser, D. Ghoshdastidar. Analysis of convolutions, non-linearity and depth in graph neural networks using neural tangent kernel. Transactions of Machine Learning Research, 2023. [Paper] [Preprint] [Code]
16. A. Mandal, M. Perrot, D. Ghoshdastidar. A revenue function for comparison-based hierarchical clustering. Transactions of Machine Learning Research, 2023 [Paper] [Preprint]
17. P. Esser, S. Mukherjee, M. Sabanayagam, D. Ghoshdastidar. Improved representation learning through tensorized autoencoders. AISTATS 2023. [Paper] [Preprint]
18. M. C. May, Z. Fang, M. Eitel, N. Stricker, D. Ghoshdastidar, G. Lanza. Graph-based prediction of missing KPIs through optimization and random forests for KPI Systems. Production Engineering, 2022. DOI:10.1007/s11740-022-01179-y [Paper]
19. L. Vankadara, L. Rendsburg, U. von Luxburg, D. Ghoshdastidar. Interpolation and regularization for causal learning. NeurIPS 2022 [Paper] [Preprint]
20. L. C. Vankadara, P. M. Faller, M. Hardt, L. Minorics, D. Ghoshdastidar, D. Janzing. Causal forecasting: Generalization bounds for autoregressive models. UAI 2022. [Paper] [Preprint] [Code]
21. M. Sabanayagam, L. C. Vankadara, D. Ghoshdastidar. Graphon based clustering and testing of networks: Algorithms and theory. ICLR 2022. [Paper] [Preprint] [Code]
22. S. Huber, S. H. Suyu, D. Ghoshdastidar, S. Taubenberger, V. Bonvin, J. H. H. Chan, M. Kromer, U. M. Noebauer, S. A. Sim, L. Leal-Taixé. HOLISMOKES -- VII. Time-delay measurement of strongly lensed Type Ia supernovae using machine learning. Astronomy & Astrophysics 2022. [Paper] [Preprint]
23. P. Esser, L. C. Vankadara, D. Ghoshdastidar. Learning theory can (sometimes) explain generalisation in graph neural networks. NeurIPS 2021 [Paper] [Preprint]
24. L. C. Vankadara, S. Bordt, U. von Luxburg, D. Ghoshdastidar. Recovery Guarantees for Kernel-based Clustering under Non-parametric Mixture Models. AISTATS 2021 [Paper] (oral presentation, top 3%)
25. M. Perrot, P. Esser, D. Ghoshdastidar. Near-optimal comparison based clustering. Neurips 2020 [Paper] [Preprint] [Code]
26. D. Ghoshdastidar, M. Gutzeit, A. Carpentier, U. Von Luxburg. Two-sample hypothesis testing for inhomogeneous random graphs. The Annals of Statistics, 48 (4): 2208-2229, 2020. [Paper] [Preprint]
27. L. C. Vankadara, D. Ghoshdastidar. On the optimality of kernels for high-dimensional clustering. AISTATS 2020. [Paper] [Preprint] [Video]
28. D. Ghoshdastidar, M. Perrot, U. Von Luxburg. Foundations of comparison-based hierarchical clustering. NeurIPS, 2019. [Paper] [Preprint] [Code]
29. D. Ghoshdastidar, U. Von Luxburg. Practical methods for graph two-sample testing. NeurIPS, 2018. [Paper] [Preprint] [Code]
30. D. Ghoshdastidar, M. Gutzeit, A. Carpentier, U. von Luxburg. Two-sample tests for large random graphs using network statistics. COLT, 2017. [Preprint]
31. D. Ghoshdastidar, A. Dukkipati. Uniform hypergraph partitioning: Provable tensor methods and sampling techniques. The Journal of Machine Learning Research, 18(50): 1-41, 2017. [Paper] [Preprint] [Code]
32. S. Haghiri, D. Ghoshdastidar, U. von Luxburg. Comparison based nearest neighbor search. AISTATS, 2017. [Paper] [Preprint]
33. D. Ghoshdastidar, A. Dukkipati. Consistency of Spectral Hypergraph Partitioning under Planted Partition Model. The Annals of Statistics, 45 (1): 289-315, 2017. [Paper] [Preprint]
34. A. Dukkipati, D. Ghoshdastidar, J. Krishnan. Mixture modelling with compact support distributions for unsupervised learning. IJCNN, 2016. [Paper]
35. D. Ghoshdastidar, A. P. Adsul, A. Dukkipati. Learning with Jensen-Tsallis kernels. IEEE Transactions on Neural Networks and Learning Systems, 27 (10), pp. 2108-2119, 2016. [Paper]
36. D. Ghoshdastidar, A. Dukkipati. A provable generalized tensor spectral method for uniform hypergraph partitioning. ICML, 2015. [Paper] [Video]
37. D. Ghoshdastidar, A. Dukkipati. Spectral clustering using multilinear SVD: Analysis, approximations and applications. AAAI, 2015. [Paper] (oral presentation, top 11%)
38. D. Ghoshdastidar, A. Dukkipati. Consistency of spectral partitioning of uniform hypergraphs under planted partition model. NIPS, 2014. [Paper]
39. D. Ghoshdastidar, A. Dukkipati, S. Bhatnagar. Newton based stochastic optimization using q-Gaussian smoothed functional algorithms. Automatica, 50(10): 2606-2614, 2014. [Paper] [Preprint]
40. D. Ghoshdastidar, A. Dukkipati, A. P. Adsul, A. S. Vijayan. Spectral clustering with Jensen-type kernels and their multi-point extensions. CVPR, 2014. [Paper] [Preprint]
41. D. Ghoshdastidar, A. Dukkipati, S. Bhatnagar. Smoothed functional algorithms for stochastic optimization using q-Gaussian distributions. ACM Transactions on Modeling and Computer Simulation, 24(3): Article 17, 2014. [Paper] [Preprint]
42. A. Dukkipati, G. Pandey, D. Ghoshdastidar, P. Koley, D. M. V. Satya Sriram. Generative maximum entropy learning for multiclass classification. ICDM, 2013. [Paper] [Preprint]
43. D. Ghoshdastidar, A. Dukkipati. On power law kernels, corresponding Reproducing Kernel Hilbert Space and applications. AAAI, 2013. [Paper]
44. D. Ghoshdastidar, A. Dukkipati, S. Bhatnagar. q-Gaussian based Smoothed Functional algorithms for stochastic optimization. ISIT, 2012. [Paper] [Preprint]

Doctoral dissertations

1. Mahalakshmi Sabanayagam. Robust machine learning from the theory of overparameterization: Neural tangents and tensorization. Technical University of Munich, 2025. [PDF]
2. Pascal Mattia Esser. Theoretical foundations for exploiting unlabelled data in machine learning. Technical University of Munich, 2024. [PDF]
3. Leena Chennuru Vankadara. Towards a theory of learning under extreme non-identifiability: Through the lens of causal learning and kernel clustering. University of Tübingen, 2023. (Wilhelm Schickard Dissertation Award) [PDF]
4. Debarghya Ghoshdastidar. Consistency of spectral algorithms for hypergraphs under planted partition model. Indian Institute of Science, 2017. (Nominated for ACM India Doctoral Dissertation Award) [PDF]

Workshop papers / Preprints

1. A. Craciun. Linear independence of powers for polynomials. [Preprint]
2. M. Sabanayagam, O. Al-Dabooni, P. Esser. Cluster specific representation learning. [Preprint]
3. L. Gosch, M. Sabanayagam, D. Ghoshdastidar, S. Günnemann. Provable robustness of (graph) neural networks against data poisoning and backdoor attacks. Neurips 2024 Workshop on New Frontiers in Adversarial Machine Learning (AdvML-Frontiers) [Preprint] (best paper award)
4. V. Kaufmann, P. M. Esser, B. Zhu, D. Ghoshdatidar. Self-supervised learning approaches to improve residential housing energy prediction accuracy. GNI Symposium & Expo on Artificial Intelligence for the Built World 2024.
5. O. Bouattour, O. Kanjilal, D. Ghoshdastidar. Active learning surrogates for enhanced reliability assessment of engineering systems. GNI Symposium & Expo on Artificial Intelligence for the Built World 2024.
6. M. Eppert, S. Mukherjee, D. Ghoshdastidar. Recovering imbalanced clusters via gradient based projection pursuit. LinStat 2024
7. M. Sabanayagam, F. Behrens, U. Adomaityte, A. Dawid. Unveiling the Hessian’s connection to the decision boundary. Mathematics of Modern Machine Learning Workshop, NeurIPS 2023 [Preprint]
8. M. Sabanayagam, P. Esser, D. Ghoshdastidar. New insights into graph convolutional networks using neural tangent kernels. MLG2022@ECMLPKDD  [Preprint]
9. N. Ayday, D. Ghoshdastidar. Improvement on incremental spectral clustering. LWDA 2021.
10. V. Starlinger, C. de la Rua Lope and D. Ghoshdastidar. Machine Learning Benchmark to Assess the Environmental Impact of Cars. AAAI 2021 AI for Urban Mobility Workshop. [Data/code]
11. D. Ghoshdastidar and U. von Luxburg. Do nonparametric two-sample tests work for small sample size? A study on random graphs. NIPS-2016 workshop on Adaptive and Scalable Nonparametric Methods in ML.
12. D. Ghoshdastidar, A. Dukkipati. Coloring random non-uniform bipartite hypergraphs. [Preprint]