Publications

NeurIPS 2026

Reasoning as Compression: Unifying Budget Forcing via the Conditional Information Bottleneck

Fabio Valerio Massoli, Andrey Kuzmin, Arash Behboodi

Treats chain-of-thought as lossy compression. A conditional information bottleneck yields an RL objective that removes redundant reasoning tokens and keeps the ones that matter, with length penalties as a special case.

arXiv 2026

When Do Stored Facts Become Available in Looped Transformers?

Fabio Valerio Massoli et al.

Tracks at which loop iteration memorized facts become usable, separating recall from knowledge use across depth.

arXiv 2026

Enabling Dynamic Computation in Looped LMs

Aayush Mishra, Arnau Padrés Masdemont, Victor Conchello Vendrell, Jordi Ros Giralt, Arash Behboodi, Fabio Valerio Massoli

Per-token adaptive depth for looped language models: easy tokens leave the loop early, hard tokens keep iterating.

arXiv 2026

Memory-Efficient Looped Transformer: Decoupling Compute from Memory in Looped Language Models

Victor Conchello Vendrell, Arnau Padres Masdemont, Niccolò Grillo, Jordi Ros-Giralt, Arash Behboodi, Fabio Valerio Massoli

Looped models trade parameters for compute, but their KV cache grows with every loop. Decoupling the two cuts the KV footprint of Ouro models by 4x (32 GB to 8 GB).

White paper + NeurIPS 2025 demo 2026

Efficient Reasoning on the Edge

Yelysei Bondarenko, Thomas Hehn, Rob Hesselink, Romain Lepert, Fabio Valerio Massoli, Evgeny Mironov, Leyla Mirvakhabova, Tribhuvanesh Orekondy, Spyridon Stasis, Andrey Kuzmin, Anna Kuzina, Markus Nagel, Ankita Nayak, Corrado Rainone, Ork de Rooij, Paul N Whatmough, Arash Behboodi, Babak Ehteshami Bejnordi

Reasoning LLMs running within on-device budgets. Core contributor: RL pipelines that shorten chains of thought by up to 5x.

ACL Findings 2026

LUMINA: Long-horizon Understanding for Multi-turn Interactive Agents

Amin Rakhsha, Thomas Hehn, Pietro Mazzaglia, Fabio Valerio Massoli, Arash Behboodi, Tribhuvanesh Orekondy

Preprint 2025

Fundamental bounds on efficiency-confidence trade-off for transductive conformal prediction

Arash Behboodi, Alvaro H.C. Correia, Fabio Valerio Massoli, Christos Louizos

NeurIPS 2024

An Information Theoretic Perspective on Conformal Prediction

Alvaro H.C. Correia, Fabio Valerio Massoli, Christos Louizos, Arash Behboodi

Links conformal prediction to information theory, giving upper bounds on intrinsic uncertainty and new training objectives for conformal models.

Preprint 2024

Simulating, Fast and Slow: Learning Policies for Black-Box Optimization

Fabio Valerio Massoli, Tim Bakker, Thomas Hehn, Tribhuvanesh Orekondy, Arash Behboodi

A learned policy decides when to call an expensive simulator and when to trust a fast differentiable surrogate, cutting simulator calls by up to about 90% on black-box optimization tasks.

TMLR 2024

Variational Learning ISTA

Fabio Valerio Massoli, Christos Louizos, Arash Behboodi

IEEE TNNLS 2022

MOCCA: Multilayer One-Class Classification for Anomaly Detection

Fabio Valerio Massoli, Fabrizio Falchi, Alperen Kantarci, Şeymanur Akti, Hazim Kemal Ekenel, Giuseppe Amato

Anomaly detection that uses every layer, not just the last: each layer's features are trained to cluster normal data, and their distances are combined at test time.

ICML 2022

Equivariant Priors for Compressed Sensing with Unknown Orientation

Anna Kuzina, Kumar Pratik, Fabio Valerio Massoli, Arash Behboodi

CVIU 2021

Detection of Face Recognition Adversarial Attacks

Fabio Valerio Massoli, Fabio Carrara, Giuseppe Amato, Fabrizio Falchi