Publications

Towards Optimizing SQL Generation via LLM Routing
Mohammadhossein Malekpour
Nour Shaheen
Foutse Khomh
Amine Mhedhbi
Text-to-SQL enables users to interact with databases through natural language, simplifying access to structured data. Although highly capabl… (see more)e large language models (LLMs) achieve strong accuracy for complex queries, they incur unnecessary latency and dollar cost for simpler ones. In this paper, we introduce the first LLM routing approach for Text-to-SQL, which dynamically selects the most cost-effective LLM capable of generating accurate SQL for each query. We present two routing strategies (score- and classification-based) that achieve accuracy comparable to the most capable LLM while reducing costs. We design the routers for ease of training and efficient inference. In our experiments, we highlight a practical and explainable accuracy-cost trade-off on the BIRD dataset.
2024-11-06
ArXiv (preprint)
arxiv.org
arxiv.org
GitChameleon: Unmasking the Version-Switching Capabilities of Code Generation Models
Nizar Islah
Justine Gehring
Diganta Misra
Eilif Muller
Irina Rish
Terry Yue Zhuo
Massimo Caccia
2024-11-05
ArXiv (preprint)
arxiv.org
arxiv.org
Imagining and building wise machines: The centrality of AI metacognition
Samuel G. B. Johnson
Amir-Hossein Karimi
Yoshua Bengio
Nick Chater
Tobias Gerstenberg
Kate Larson
Sydney Levine
Melanie Mitchell
Iyad Rahwan
Bernhard Schölkopf
Igor Grossmann
2024-11-04
ArXiv (preprint)
arxiv.org
arxiv.org
Seq-VCR: Preventing Collapse in Intermediate Transformer Representations for Enhanced Reasoning
Md Rifat Arefin
Gopeshh Raaj Subbaraj
Nicolas Gontier
Yann LeCun
Irina Rish
Ravid Shwartz-Ziv
Chris Pal
2024-11-04
ArXiv (preprint)
arxiv.org
arxiv.org
Crystal Design Amidst Noisy DFT Signals: A Reinforcement Learning Approach
Prashant Govindarajan
Mathieu Reymond
Santiago Miret
Mariano Phielipp
Sarath Chandar
2024-11-03
NeurIPS.cc/2024/Workshop/AI4Mat (published)
openreview.net
openreview.net
Molphenix: A Multimodal Foundation Model for PhenoMolecular Retrieval
Philip Fradkin
Puria Azadi Moghadam
Karush Suri
Frederik Wenkel
Maciej Sypetkowski
Dominique Beaini
Predicting molecular impact on cellular function is a core challenge in therapeutic design. Phenomic experiments, designed to capture cellu… (see more)lar morphology, utilize microscopy based techniques and demonstrate a high throughput solution for uncovering molecular impact on the cell. In this work, we learn a joint latent space between molecular structures and microscopy phenomic experiments, aligning paired samples with contrastive learning. Specifically, we study the problem of Contrastive PhenoMolecular Retrieval, which consists of zero-shot molecular structure identification conditioned on phenomic experiments. We assess challenges in multi-modal learning of phenomics and molecular modalities such as experimental batch effect, inactive molecule perturbations, and encoding perturbation concentration. We demonstrate improved multi-modal learner retrieval through (1) a uni-modal pre-trained phenomics model, (2) a novel inter sample similarity aware loss, and (3) models conditioned on a representation of molecular concentration. Following this recipe, we propose MolPhenix, a molecular phenomics model. MolPhenix leverages a pre-trained phenomics model to demonstrate significant performance gains across perturbation concentrations, molecular scaffolds, and activity thresholds. In particular, we demonstrate an 8.1
2024-11-02
NeurIPS.cc/2024/Workshop/FM4Science (oral)
openreview.net
openreview.net
Trained Without My Consent: Detecting Code Inclusion In Language Models Trained on Code
Vahid Majdinasab
Amin Nikanjam
Foutse Khomh
Code auditing ensures that the developed code adheres to standards, regulations, and copyright protection by verifying that it does not cont… (see more)ain code from protected sources. The recent advent of Large Language Models (LLMs) as coding assistants in the software development process poses new challenges for code auditing. The dataset for training these models is mainly collected from publicly available sources. This raises the issue of intellectual property infringement as developers' codes are already included in the dataset. Therefore, auditing code developed using LLMs is challenging, as it is difficult to reliably assert if an LLM used during development has been trained on specific copyrighted codes, given that we do not have access to the training datasets of these models. Given the non-disclosure of the training datasets, traditional approaches such as code clone detection are insufficient for asserting copyright infringement. To address this challenge, we propose a new approach, TraWiC; a model-agnostic and interpretable method based on membership inference for detecting code inclusion in an LLM's training dataset. We extract syntactic and semantic identifiers unique to each program to train a classifier for detecting code inclusion. In our experiments, we observe that TraWiC is capable of detecting 83.87% of codes that were used to train an LLM. In comparison, the prevalent clone detection tool NiCad is only capable of detecting 47.64%. In addition to its remarkable performance, TraWiC has low resource overhead in contrast to pair-wise clone detection that is conducted during the auditing process of tools like CodeWhisperer reference tracker, across thousands of code snippets.
2024-11-02
ACM Transactions on Software Engineering and Methodology (published)
doi.org
arxiv.org
Community-based reconstruction and simulation of a full-scale model of the rat hippocampus CA1 region
Armando Romani
Alberto Antonietti
Davide Bella
Julian Budd
Elisabetta Giacalone
Kerem Kurban
Sára Sáray
Marwan Abdellah
Alexis Arnaudon
Elvis Boci
Cristina Colangelo
Jean-Denis Courcol
Thomas Delemontex
András Ecker
Joanne Falck
Cyrille Favreau
Michael Gevaert
Juan B. Hernando
Joni Herttuainen
Genrich Ivaska … (see 28 more)
Lida Kanari
Anna-Kristin Kaufmann
James King
Pramod Kumbhar
Sigrun Lange
Huanxiang Lu
Carmen Alina Lupascu
Rosanna Migliore
Fabien Petitjean
Judit Planas
Pranav Rai
Srikanth Ramaswamy
Michael W. Reimann
Juan Luis Riquelme
Nadir Román Guerrero
Ying Shi
Vishal Sood
Mohameth François Sy
Werner Van Geit
Liesbeth Vanherpe
Tamás F. Freund
Audrey Mercer
Eilif Benjamin Muller
Felix Schürmann
Alex M. Thomson
Michele Migliore
Szabolcs Káli
Henry Markram
The CA1 region of the hippocampus is one of the most studied regions of the rodent brain, thought to play an important role in cognitive fun… (see more)ctions such as memory and spatial navigation. Despite a wealth of experimental data on its structure and function, it has been challenging to integrate information obtained from diverse experimental approaches. To address this challenge, we present a community-based, full-scale in silico model of the rat CA1 that integrates a broad range of experimental data, from synapse to network, including the reconstruction of its principal afferents, the Schaffer collaterals, and a model of the effects that acetylcholine has on the system. We tested and validated each model component and the final network model, and made input data, assumptions, and strategies explicit and transparent. The unique flexibility of the model allows scientists to potentially address a range of scientific questions. In this article, we describe the methods used to set up simulations to reproduce in vitro and in vivo experiments. Among several applications in the article, we focus on theta rhythm, a prominent hippocampal oscillation associated with various behavioral correlates and use our computer model to reproduce experimental findings. Finally, we make data, code, and model available through the hippocampushub.eu portal, which also provides an extensive set of analyses of the model and a user-friendly interface to facilitate adoption and usage. This community-based model represents a valuable tool for integrating diverse experimental data and provides a foundation for further research into the complex workings of the hippocampal CA1 region.
2024-11-01
PLoS Biology (published)
doi.org
A protocol for trustworthy EEG decoding with neural networks
Davide Borra
Elisa Magosso
Mirco Ravanelli
2024-11-01
Neural Networks (published)
doi.org
SpeechBrain-MOABB: An open-source Python library for benchmarking deep neural networks applied to EEG signals
Davide Borra
Francesco Paissan
Mirco Ravanelli
2024-11-01
Computers in Biology and Medicine (published)
doi.org
Structured Pruning of Neural Networks for Constraints Learning
Matteo Cacciola
Antonio Frangioni
Andrea Lodi
In recent years, the integration of Machine Learning (ML) models with Operation Research (OR) tools has gained popularity across diverse app… (see more)lications, including cancer treatment, algorithmic configuration, and chemical process optimization. In this domain, the combination of ML and OR often relies on representing the ML model output using Mixed Integer Programming (MIP) formulations. Numerous studies in the literature have developed such formulations for many ML predictors, with a particular emphasis on Artificial Neural Networks (ANNs) due to their significant interest in many applications. However, ANNs frequently contain a large number of parameters, resulting in MIP formulations that are impractical to solve, thereby impeding scalability. In fact, the ML community has already introduced several techniques to reduce the parameter count of ANNs without compromising their performance, since the substantial size of modern ANNs presents challenges for ML applications as it significantly impacts computational efforts during training and necessitates significant memory resources for storage. In this paper, we showcase the effectiveness of pruning, one of these techniques, when applied to ANNs prior to their integration into MIPs. By pruning the ANN, we achieve significant improvements in the speed of the solution process. We discuss why pruning is more suitable in this context compared to other ML compression techniques, and we identify the most appropriate pruning strategies. To highlight the potential of this approach, we conduct experiments using feed-forward neural networks with multiple layers to construct adversarial examples. Our results demonstrate that pruning offers remarkable reductions in solution times without hindering the quality of the final decision, enabling the resolution of previously unsolvable instances.
2024-11-01
Operations Research Letters (published)
doi.org
arxiv.org
Soft Condorcet Optimization for Ranking of General Agents
Marc Lanctot
Kate Larson
Michael Kaisers
Quentin Berthet
Ian Gemp
Manfred Diaz
Roberto-Rafael Maura-Rivero
Yoram Bachrach
Anna Koop
Doina Precup
A common way to drive progress of AI models and agents is to compare their performance on standardized benchmarks. Comparing the performance… (see more) of general agents requires aggregating their individual performances across a potentially wide variety of different tasks. In this paper, we describe a novel ranking scheme inspired by social choice frameworks, called Soft Condorcet Optimization (SCO), to compute the optimal ranking of agents: the one that makes the fewest mistakes in predicting the agent comparisons in the evaluation data. This optimal ranking is the maximum likelihood estimate when evaluation data (which we view as votes) are interpreted as noisy samples from a ground truth ranking, a solution to Condorcet's original voting system criteria. SCO ratings are maximal for Condorcet winners when they exist, which we show is not necessarily true for the classical rating system Elo. We propose three optimization algorithms to compute SCO ratings and evaluate their empirical performance. When serving as an approximation to the Kemeny-Young voting method, SCO rankings are on average 0 to 0.043 away from the optimal ranking in normalized Kendall-tau distance across 865 preference profiles from the PrefLib open ranking archive. In a simulated noisy tournament setting, SCO achieves accurate approximations to the ground truth ranking and the best among several baselines when 59\% or more of the preference data is missing. Finally, SCO ranking provides the best approximation to the optimal ranking, measured on held-out test sets, in a problem containing 52,958 human players across 31,049 games of the classic seven-player game of Diplomacy.
2024-10-31
ArXiv (preprint)
arxiv.org
arxiv.org