Abbas Rahimi

Research Interests:

My primary research interest lies in creating formally expressive machine learning and reasoning algorithms, yet with scalable and high-performance implementations on classical and quantum computing platforms. This includes AI agents based on hybrid foundation models, with the main emphasis on improving the quality and cost of AI-driven design automation, optimization, and discovery. Additionally, I am broadly interested in exploiting cross-layer approximation opportunities spanning computation, communication, sensing, and storage systems to significantly reduce computational complexity and energy consumption.

Short Biography:

I joined IBM Research–Zürich in 2020, where I am currently a Research Staff Member in the Mathematics of Computation Department. From 2015 to 2020, I was a postdoctoral researcher at UC Berkeley and a Postdoctoral Fellow at ETH Zürich. I received my M.S. and Ph.D. in Computer Science and Engineering from UC San Diego in 2015, and my B.S. in Computer Engineering from the University of Tehran in 2010.

I received the 2015 Outstanding Dissertation Award from the European Design and Automation Association in the area of New Directions in Embedded System Design and Embedded Software, as well as the ETH Zürich Postdoctoral Fellowship in 2017. My work has earned Best Paper Awards at BICT (2017), BioCAS (2018), and IBM’s Pat Goldberg Memorial Award (2020), along with Best Paper Nominations at DAC (2013) and DATE (2019), and Spotlight Papers at NeurIPS (2025 and 2026). I have also been honored with IBM’s Outstanding Technical Achievement Award in both 2023 and 2025. In addition, my research was highlighted by Quanta Magazine as one of the major breakthroughs in computer science in 2023.

Selected Recent Publications:

• N. Menet, M. Hersche, A. Krause, A. Rahimi, A theoretical analysis of test-driven code generation, Conference on Neural Information Processing Systems (NeurIPS), 2026. (Spotlight paper)

• M. Hersche, N. Menet, R. Tanios, A. Rahimi, Locally coherent parallel decoding in diffusion language models, The International Conference on Machine Learning (ICML), 2026.

• M. Hersche, S. Moor-Smith, T. Hofmann, A. Rahimi, Soft-masked diffusion language models, International Conference on Learning Representations (ICLR), 2026.

• F. Carzaniga, M. Hersche, A. Sebastian, K. Schindler, A. Rahimi, A foundation model with multi-variate parallel attention to generate neuronal activity, International Conference on Learning Representations (ICLR), 2026. [Dataset] [Code] [checkpoints]

• A. Terzic, N. Menet, M. Hersche, T. Hofmann, A. Rahimi, Structured sparse transition matrices to enable state tracking in state-space models, Conference on Neural Information Processing Systems (NeurIPS), 2025. (Spotlight paper)

• G. Camposampiero, P. Barbiero, M. Hersche, R. Wattenhofer, A. Rahimi, Scalable evaluation and neural models for compositional generalization, Conference on Neural Information Processing Systems (NeurIPS), 2025.

• F. Carzaniga, G. Hoppeler, M. Hersche, K. Schindler, A. Rahimi, The case for cleaner biosignals: high-fidelity neural compressor enables transfer from cleaner iEEG to noisier EEG, International Conference on Learning Representations (ICLR), 2025.

• J. Büchel, A. Vasilopoulos, W. Simon, I. Boybat, H. Tsai, G. Burr, H. Castro, B. Filipiak, M. Le Gallo, A. Rahimi, V. Narayanan, A. Sebastian, Efficient scaling of large language models with mixture of experts and 3D analog in-memory computing, Nature Computational Science, 2025. (Featured on the cover Jan issue 2025. Highlighted in Nat. Comput. Sci's News & Views)

• J. Thomm, G. Camposampiero, A. Terzic, M. Hersche, B. Schölkopf, A. Rahimi, Limits of transformer language models on learning to compose algorithms, Conference on Neural Information Processing Systems (NeurIPS), 2024.

• J. Büchel, G. Camposampiero, A. Vasilopoulos, C. Lammie, M. Le Gallo, A. Rahimi, A. Sebastian, Kernel approximation using analog in-memory computing, Nature Machine Intelligence, 2024.

• N. Menet, M. Hersche, K. Karunaratne, L. Benini, A. Sebastian, A. Rahimi, MIMONets: multiple-input-multiple-output neural networks exploiting computation in superposition Conference on Neural Information Processing Systems (NeurIPS), 2023.

• M. Hersche, M. Zeqiri, L. Benini, A. Sebastian, A. Rahimi, A neuro-vector-symbolic architecture for solving Raven’s progressive matrices, Nature Machine Intelligence, 2023. (Spotlighted by Quanta Magazine as one of 2023's biggest breakthroughs in CS)

• J. Langenegger, G. Karunaratne, M. Hersche, L. Benini, A. Sebastian, A. Rahimi, In-memory factorization of holographic perceptual representations, Nature Nanotechnology, 2023. (Highlighted in Nat. Nanotechnol.'s News & Views)

• M. Hersche, G. Karunaratne, G. Cherubini, L. Benini, A. Sebastian, A. Rahimi, Constrained few-shot class-incremental learning,  Conference on Computer Vision and Pattern Recognition (CVPR), 2022.

• G. Karunaratne, M. Schmuck, M. Le Gallo, G. Cherubini, L. Benini, A. Sebastian, A. Rahimi, Robust high-dimensional memory-augmented neural networks, Nature Communications, 2021. (Featured in the 50 best articles in the Nat. Comm's Applied Physics and Mathematics)

• A. Moin, A. Zhou, A. Rahimi, et al., A wearable biosensing system with in-sensor adaptive machine learning for hand gesture recognition, Nature Electronics, 2021. (Highlighted in Nat. Electron.'s News & Views)

• G. Karunaratne, M. Le Gallo, G. Cherubini, L. Benini, A. Rahimi, A. Sebastian, In-memory hyperdimensional computing, Nature Electronics, 2020. (Featured on the cover June issue 2020. Received IBM's Pat Goldberg Memorial Best Paper Awards)

Research in the News:

• The quest to teach LLMs how to count

• 2023's Biggest Breakthroughs in Computer Science

• A New Approach to Computation Reimagines Artificial Intelligence

• Disentangling visual concepts by embracing stochastic in-memory computing

• Mimicking the brain: Deep learning meets vector-symbolic AI

• The best of both worlds: Deep learning meets vector-symbolic architectures

• High-five or thumbs-up? New device detects which hand gesture you want to make

• Fulfilling Brain-inspired Hyperdimensional Computing with In-memory Computing