🧑‍🔬 Research

📚 Publications

Conference Proceedings

[1] J. Mukhoti^*, A. Kirsch^*, J. van Amersfoort, P. H. Torr, and Y. Gal, "Deterministic Neural Networks with Appropriate Inductive Biases Capture Epistemic and Aleatoric Uncertainty," CVPR 2023, 2023.

[2] F. Bickford Smith^*, A. Kirsch^*, S. Farquhar, Y. Gal, A. Foster, and T. Rainforth, "Prediction-Oriented Bayesian Active Learning," AISTATS, 2023.

[3] S. Mindermann^*, J. M. Brauner^*, M. T. Razzak^*, A. Kirsch, et al., "Prioritized Training on Points that are Learnable, Worth Learning, and not yet Learnt," ICML, 2022.

[4] A. Jesson^*, P. Tigas^*, J. van Amersfoort, A. Kirsch, U. Shalit, and Y. Gal, "Causal-BALD: Deep Bayesian Active Learning of Outcomes to Infer Treatment-Effects from Observational Data," NeurIPS, 2021.

[5] A. Kirsch^*, J. van Amersfoort^*, and Y. Gal, "BatchBALD: Efficient and Diverse Batch Acquisition for Deep Bayesian Active Learning," NeurIPS, 2019.

Journal Articles

[6] A. Kirsch, "Black-Box Batch Active Learning for Regression", TMLR, 2023.

[7] A. Kirsch, "Does ‘Deep Learning on a Data Diet’ reproduce? Overall yes, but GraNd at Initialization does not", TMLR, 2023.

[8] A. Kirsch^*, S. Farquhar^*, P. Atighehchian, A. Jesson, F. Branchaud-Charron, Y. Gal, "Stochastic Batch Acquisition: A Simple Baseline for Deep Active Learning", TMLR, 2023.

[9] A. Kirsch and Y. Gal, "A Note on "Assessing Generalization of SGD via Disagreement"," TMLR, 2022.

[10] A. Kirsch and Y. Gal, "Unifying Approaches in Data Subset Selection via Fisher Information and Information-Theoretic Quantities," TMLR, 2022.

Workshop Papers

[11] D. Tran, J. Liu, M. W. Dusenberry, et al., "Plex: Towards Reliability using Pretrained Large Model Extensions," Principles of Distribution Shifts & First Workshop on Pre-training: Perspectives, Pitfalls, and Paths Forward, ICML 2022.

[12] A. Kirsch, J. Kossen, and Y. Gal, "Marginal and Joint Cross-Entropies & Predictives for Online Bayesian Inference, Active Learning, and Active Sampling," Updatable Machine Learning, ICML 2022, 2022.

[13] A. Kirsch, J. Mukhoti, J. van Amersfoort, P. H. Torr, and Y. Gal, "On Pitfalls in OoD Detection: Entropy Considered Harmful," Uncertainty in Deep Learning, 2021.

[14] A. Kirsch, T. Rainforth, and Y. Gal, "Active Learning under Pool Set Distribution Shift and Noisy Data," SubSetML, 2021.

[15] A. Kirsch^*, S. Farquhar^*, and Y. Gal, "A Simple Baseline for Batch Active Learning with Stochastic Acquisition Functions," SubSetML, 2021.

[16] A. Kirsch and Y. Gal, "A Practical & Unified Notation for Information-Theoretic Quantities in ML," SubSetML, 2021.

[17] A. Kirsch, C. Lyle, and Y. Gal, "Scalable Training with Information Bottleneck Objectives," Uncertainty in Deep Learning,

[18] A. Kirsch, C. Lyle, and Y. Gal, "Learning CIFAR-10 with a Simple Entropy Estimator Using Information Bottleneck Objectives," Uncertainty in Deep Learning, 2020.

---

📝 Reviewing

NeurIPS 2019 (Top Reviewer), AAAI 2020, AAAI 2021, ICLR 2021, NeurIPS 2021 (Outstanding Reviewer), NeurIPS 2022, NeurIPS 2022, TMLR, CVPR 2023.