🔥 News

• Jun 2026
  🚀 Coming soon: “KVarN: Variance-Normalized KV-Cache Quantization Mitigates Error Accumulation in Reasoning Tasks”. Stay tuned!
• May 2026
  🎉 My paper “SINQ: Sinkhorn-Normalized Quantization for Calibration-Free Low-Precision LLM Weights” has been accepted at ICML 2026!
• Feb 2026
  🤗 SINQ is now integrated into Hugging Face Transformers! Check out the code and docs on GitHub.

📝 Selected Publications

• Coming soon
  KVarN: Variance-Normalized KV-Cache Quantization Mitigates Error Accumulation in Reasoning Tasks Lorenz K. Mueller, Philippe Bich, Chiara Boretti, Hyun Min Chang, Jiawei Zhuang, Lukas Cavigelli Preprint, 2026.

  TL;DR: A novel state-of-the-art variance-normalized KV-cache quantization scheme that limits compounding error in long reasoning traces and beats TurboQuant by Google Research with better accuracy and lower bits.

• ICML 2026
  SINQ: Sinkhorn-Normalized Quantization for Calibration-Free Low-Precision LLM Weights Lorenz K. Mueller, Philippe Bich, Jiawei Zhuang, Ahmet Çalik, Luca Benfenati, Lukas Cavigelli International Conference on Machine Learning (ICML), 2026.

  TL;DR: A calibration-free quantization method based on Sinkhorn normalization that delivers strong low-bit LLM weights out of the box. Integrated into 🤗 Hugging Face Transformers.

• IEEE TPAMI 2025
  On the Universal Approximation Properties of Deep Neural Networks using MAM Neurons Philippe Bich, Andriy Enttsel, Luciano Prono, Alex Marchioni, Fabio Pareschi, Mauro Mangia, Gianluca Setti, Riccardo Rovatti IEEE Transactions on Pattern Analysis and Machine Intelligence, 2025.

  TL;DR: Theoretical foundations showing that Multiply-And-Max/min (MAM) neurons preserve the universal approximation property while enabling aggressive structured pruning.

Browse all publications

📖 Education

• 
  Politecnico di Torino

  • Ph.D. in Electrical, Electronics and Communications Engineering Nov 2021 – 2025 · Advisor: Prof. Gianluca Setti · AI model compression and quantization
  • M.Sc. in Mechatronics Engineering — 110/110 cum laude 2018 – 2021 · Master's thesis at the Boston University Robotics Lab with Prof. John Baillieul
  • B.Sc. in Computer Engineering — 109/110 2015 – 2018

📚 All Publications

• ICML
  SINQ: Sinkhorn-Normalized Quantization for Calibration-Free Low-Precision LLM Weights International Conference on Machine Learning (ICML), 2026.
• J-STARS
  FOREST-GC: A conFOrmable Rendering Engine for Synthetic Tree Generation and Counting IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing, 2026.

• TNNLS
  A Multiply-And-Max/min Neuron Paradigm for Aggressively Prunable Deep Neural Networks IEEE Transactions on Neural Networks and Learning Systems, vol. 36, no. 8, pp. 14414–14427, 2025.
• TPAMI
  On the Universal Approximation Properties of Deep Neural Networks using MAM Neurons IEEE Transactions on Pattern Analysis and Machine Intelligence, 2025.
• MLJ
  Linearly-Interpretable Concept Embedding Models for Text Analysis Machine Learning, vol. 114, no. 10, art. 224, 2025.
• xAI
  V-CEM: Bridging Performance and Intervenability in Concept-based Models World Conference on Explainable Artificial Intelligence (xAI), pp. 48–67, 2025.
• ECML PKDD
  Towards Better Generalization and Interpretability in Unsupervised Concept-Based Models Joint European Conference on Machine Learning and Knowledge Discovery in Databases (ECML PKDD), 2025.
• TCAS-II
  MESA: A Dynamical Attention-based Pre-processing Pipeline for High-throughput Event-based Computer Vision Tasks IEEE Transactions on Circuits and Systems II: Express Briefs, 2025.

• CVPRW
  Event-based Eye Tracking: AIS 2024 Challenge Survey CVPR 2024 Workshops — AIS: Vision, Graphics and AI for Streaming.
• BioCAS
  Memory in Motion: Exploring Leaky Integration of Time Surfaces for Event-Based Eye-Tracking IEEE Biomedical Circuits and Systems Conference (BioCAS), pp. 1–5, 2024.
• AICAS
  Optimizing Vision Transformers: Leveraging Max and Min Operations for Efficient Pruning IEEE International Conference on Artificial Intelligence Circuits and Systems (AICAS), 2024.

• CVPRW
  Pedro: an Event-based Dataset for Person Detection in Robotics CVPR 2023 Workshops — 4th International Workshop on Event-Based Vision.
• MWSCAS
  Multiply-and-Max/min Neurons at the Edge: Pruned Autoencoder Implementation IEEE International Midwest Symposium on Circuits and Systems (MWSCAS), 2023.

• ICRA
  Visual Navigation Using Sparse Optical Flow and Time-to-Transit IEEE International Conference on Robotics and Automation (ICRA), pp. 9397–9403, 2022.
• BioCAS
  Aggressively Prunable MAM²-based Deep Neural Oracle for ECG Acquisition by Compressed Sensing IEEE Biomedical Circuits and Systems Conference (BioCAS), pp. 163–167, 2022.