Publications

TRUTH: Teaching LLMs to Rerank for Truth in Misinformation Detection
Hao Yu
Shenyang Huang
Zachary Yang
Maximilian Puelma Touzel
Kellin Pelrine
Jean-François Godbout
Reihaneh Rabbany
2025-07-25
colmweb.org/COLM/2025/Workshop/SoLaR (poster)
openreview.net
openreview.net
Can LLMs Reason Abstractly Over Math Word Problems Without CoT? Disentangling Abstract Formulation From Arithmetic Computation
Ziling Cheng
Meng Cao
Leila Pishdad
Yanshuai Cao
Jackie Cheung
Final-answer-based metrics are commonly used for evaluating large language models (LLMs) on math word problems, often taken as proxies for r… (see more)easoning ability. However, such metrics conflate two distinct sub-skills: abstract formulation (capturing mathematical relationships using expressions) and arithmetic computation (executing the calculations). Through a disentangled evaluation on GSM8K and SVAMP, we find that the final-answer accuracy of Llama-3 and Qwen2.5 (1B-32B) without CoT is overwhelmingly bottlenecked by the arithmetic computation step and not by the abstract formulation step. Contrary to the common belief, we show that CoT primarily aids in computation, with limited impact on abstract formulation. Mechanistically, we show that these two skills are composed conjunctively even in a single forward pass without any reasoning steps via an abstract-then-compute mechanism: models first capture problem abstractions, then handle computation. Causal patching confirms these abstractions are present, transferable, composable, and precede computation. These behavioural and mechanistic findings highlight the need for disentangled evaluation to accurately assess LLM reasoning and to guide future improvements.
2025-07-24
colmweb.org/COLM/2025/Workshop/XLLM-Reason-Plan (published)
openreview.net
openreview.net
Enhancing Logical Reasoning in Large Language Models through Graph-based Synthetic Data
Jiaming Zhou
Abbas Ghaddar
Ge Zhang
Liheng Ma
Yaochen Hu
Soumyasundar Pal
Mark Coates
Jianye HAO
B. Wang
Yingxue Zhang
2025-07-24
colmweb.org/COLM/2025/Workshop/XLLM-Reason-Plan (published)
doi.org
openreview.net
Enhancing Logical Reasoning in Large Language Models through Graph-based Synthetic Data
Jiaming Zhou
Abbas Ghaddar
Ge Zhang
Liheng Ma
Yaochen Hu
Soumyasundar Pal
B. Wang
Jianye HAO
Mark Coates
Yingxue Zhang
Despite recent advances in training and prompt- ing strategies for Large Language Models (LLMs), these models continue to face chal- lenges … (see more)with complex logical reasoning tasks that involve long reasoning chains. In this work, we explore the potential and limitations of using graph-based synthetic reasoning data as training signals to enhance LLMs’ reasoning capabilities. Our extensive experiments, con- ducted on two established natural language rea- soning tasks—inductive reasoning and spatial reasoning—demonstrate that supervised fine- tuning (SFT) with synthetic graph-based rea- soning data effectively enhances LLMs’ rea- soning performance, without compromising their effectiveness on other standard evaluation benchmarks.
2025-07-24
colmweb.org/COLM/2025/Workshop/XLLM-Reason-Plan (published)
openreview.net
openreview.net
Multiscale Neural PDE Surrogates for Prediction and Downscaling: Application to Ocean Currents
Abdessamad El-Kabid
Loubna Benabbou
Redouane Lguensat
Alex Hern'andez-Garc'ia
Accurate modeling of physical systems governed by partial differential equations is a central challenge in scientific computing. In oceanogr… (see more)aphy, high-resolution current data are critical for coastal management, environmental monitoring, and maritime safety. However, available satellite products, such as Copernicus data for sea water velocity at ~0.08 degrees spatial resolution and global ocean models, often lack the spatial granularity required for detailed local analyses. In this work, we (a) introduce a supervised deep learning framework based on neural operators for solving PDEs and providing arbitrary resolution solutions, and (b) propose downscaling models with an application to Copernicus ocean current data. Additionally, our method can model surrogate PDEs and predict solutions at arbitrary resolution, regardless of the input resolution. We evaluated our model on real-world Copernicus ocean current data and synthetic Navier-Stokes simulation datasets.
2025-07-24
ArXiv (preprint)
arxiv.org
arxiv.org
AURA: A Multi-Modal Medical Agent for Understanding, Reasoning&Annotation
Nima Fathi
Amar Kumar
Tal Arbel
Recent advancements in Large Language Models (LLMs) have catalyzed a paradigm shift from static prediction systems to agentic AI agents capa… (see more)ble of reasoning, interacting with tools, and adapting to complex tasks. While LLM-based agentic systems have shown promise across many domains, their application to medical imaging remains in its infancy. In this work, we introduce AURA, the first visual linguistic explainability agent designed specifically for comprehensive analysis, explanation, and evaluation of medical images. By enabling dynamic interactions, contextual explanations, and hypothesis testing, AURA represents a significant advancement toward more transparent, adaptable, and clinically aligned AI systems. We highlight the promise of agentic AI in transforming medical image analysis from static predictions to interactive decision support. Leveraging Qwen-32B, an LLM-based architecture, AURA integrates a modular toolbox comprising: (i) a segmentation suite with phase grounding, pathology segmentation, and anatomy segmentation to localize clinically meaningful regions; (ii) a counterfactual image-generation module that supports reasoning through image-level explanations; and (iii) a set of evaluation tools including pixel-wise difference-map analysis, classification, and advanced state-of-the-art components to assess diagnostic relevance and visual interpretability.
2025-07-22
ArXiv (preprint)
arxiv.org
arxiv.org
AURA: A Multi-Modal Medical Agent for Understanding, Reasoning&Annotation
Nima Fathi
Amar Kumar
Tal Arbel
Recent advancements in Large Language Models (LLMs) have catalyzed a paradigm shift from static prediction systems to agentic AI agents capa… (see more)ble of reasoning, interacting with tools, and adapting to complex tasks. While LLM-based agentic systems have shown promise across many domains, their application to medical imaging remains in its infancy. In this work, we introduce AURA, the first visual linguistic explainability agent designed specifically for comprehensive analysis, explanation, and evaluation of medical images. By enabling dynamic interactions, contextual explanations, and hypothesis testing, AURA represents a significant advancement toward more transparent, adaptable, and clinically aligned AI systems. We highlight the promise of agentic AI in transforming medical image analysis from static predictions to interactive decision support. Leveraging Qwen-32B, an LLM-based architecture, AURA integrates a modular toolbox comprising: (i) a segmentation suite with phase grounding, pathology segmentation, and anatomy segmentation to localize clinically meaningful regions; (ii) a counterfactual image-generation module that supports reasoning through image-level explanations; and (iii) a set of evaluation tools including pixel-wise difference-map analysis, classification, and advanced state-of-the-art components to assess diagnostic relevance and visual interpretability.
2025-07-22
ArXiv (preprint)
doi.org
arxiv.org
Penalty Learning for Optimal Partitioning using Multilayer Perceptron
Tung L. Nguyen
Toby Dylan Hocking
Changepoint detection is a technique used to identify significant shifts in sequences and is widely used in fields such as finance, genomics… (see more), and medicine. To identify the changepoints, dynamic programming (DP) algorithms, particularly Optimal Partitioning (OP) family, are widely used. To control the changepoints count, these algorithms use a fixed penalty to penalize the changepoints presence. To predict the optimal value of that penalty, existing methods used simple models such as linear or tree-based, which may limit predictive performance. To address this issue, this study proposes using a multilayer perceptron (MLP) with a ReLU activation function to predict the penalty. The proposed model generates continuous predictions -- as opposed to the stepwise ones in tree-based models -- and handles non-linearity better than linear models. Experiments on large benchmark genomic datasets demonstrate that the proposed model improves accuracy and F1 score compared to existing models.
2025-07-22
Statistics and Computing (published)
doi.org
arxiv.org
Pharmaco-nutraceutical improvement of the response to obeticholic acid with omega-3 polyunsaturated fatty acids
Audrey-Anne Lavoie
Ariane Thérien
Anisia Silva
Emanuel Paré
Anna Ciešlak
William Gagnon
Clémence Desjardins
Mélanie Verreault
Jocelyn Trottier
Marie-Claude Vohl
Jean-Philippe Drouin-Chartier
Jacques Corbeil
Alexandre Caron
Olivier Barbier
2025-07-22
Biochemical Journal (published)
doi.org
Pharmaco-nutraceutical improvement of the response to obeticholic acid with omega-3 polyunsaturated fatty acids.
Audrey-Anne Lavoie
Ariane Thérien
Anisia Silva
Emanuel Paré
Anna Ciešlak
William Gagnon
Clémence Desjardins
Mélanie Verreault
J. Trottier
Marie-Claude Vohl
Jean-Philippe Drouin-Chartier
Jacques Corbeil
Alexandre Caron
Olivier Barbier
Obeticholic acid (OCA) is the second line therapy for primary biliary cholangitis. While efficient in promoting BA detoxification and limiti… (see more)ng liver fibrosis, its clinical use is restricted by severe dose-dependent side effects. We tested the hypothesis that adding n-3 polyunsaturated fatty acids, eicosapentaenoic (EPA) and docosahexaenoic (DHA) acids to OCA may improve the therapeutic effect of the low drug dosage. Several liver cell lines were exposed to vehicle, low or high OCA dose (1-20μM) in the presence or absence of EPA/DHA for 24H. To induce ER stress, apoptosis, and fibrosis, HepG2 cells were exposed to a 400μM BA mixture or to 2ng/mL TGF-β. For inflammation analyses, THP-1 cells were activated with 100ng/mL LPS. The impact OCA+EPA/DHA was assessed using transcriptomic (qRT-PCR), proteomic (ELISA, caspase-3), and metabolomic (LC-MS/MS) approaches. The addition of EPA/DHA reinforced the ability of low OCA dose to down-regulate the expression of genes involved in BA synthesis (CYP7A1, CYP8B1) and uptake (NTCP) and to up-regulate MRP2 & 3 genes expression. EPA/DHA also enhanced the anti-inflammatory response of the drug by reducing the expression of the LPS-induced cytokines: TNFα, IL-6, IL-1β and MCP-1 in THP-1 macrophages. OCA+EPA/DHA decreased the expression of BIP, CHOP and COL1A1 genes and the caspase-3 activity. EPA+DHA potentiate the response to low OCA doses on BA toxicity, and provide additional benefits on ER stress, apoptosis, inflammation and fibrosis. These observations support the idea that adding n-3 polyunsaturated fatty acids to the drug may reduce the risk of dose-related side effects in patients treated with OCA.
2025-07-22
Biochemical Journal (published)
doi.org
Pharmaco-nutraceutical improvement of the response to obeticholic acid with omega-3 polyunsaturated fatty acids.
Audrey-Anne Lavoie
Ariane Thérien
Anisia Silva
Emanuel Paré
Anna Ciešlak
William Gagnon
Clémence Desjardins
Mélanie Verreault
Jocelyn Trottier
Marie-Claude Vohl
Jean-Philippe Drouin-Chartier
Jacques Corbeil
Alexandre Caron
Olivier Barbier
Obeticholic acid (OCA) is the second line therapy for primary biliary cholangitis. While efficient in promoting BA detoxification and limiti… (see more)ng liver fibrosis, its clinical use is restricted by severe dose-dependent side effects. We tested the hypothesis that adding n-3 polyunsaturated fatty acids, eicosapentaenoic (EPA) and docosahexaenoic (DHA) acids to OCA may improve the therapeutic effect of the low drug dosage. Several liver cell lines were exposed to vehicle, low or high OCA dose (1-20μM) in the presence or absence of EPA/DHA for 24H. To induce ER stress, apoptosis, and fibrosis, HepG2 cells were exposed to a 400μM BA mixture or to 2ng/mL TGF-β. For inflammation analyses, THP-1 cells were activated with 100ng/mL LPS. The impact OCA+EPA/DHA was assessed using transcriptomic (qRT-PCR), proteomic (ELISA, caspase-3), and metabolomic (LC-MS/MS) approaches. The addition of EPA/DHA reinforced the ability of low OCA dose to down-regulate the expression of genes involved in BA synthesis (CYP7A1, CYP8B1) and uptake (NTCP) and to up-regulate MRP2 & 3 genes expression. EPA/DHA also enhanced the anti-inflammatory response of the drug by reducing the expression of the LPS-induced cytokines: TNFα, IL-6, IL-1β and MCP-1 in THP-1 macrophages. OCA+EPA/DHA decreased the expression of BIP, CHOP and COL1A1 genes and the caspase-3 activity. EPA+DHA potentiate the response to low OCA doses on BA toxicity, and provide additional benefits on ER stress, apoptosis, inflammation and fibrosis. These observations support the idea that adding n-3 polyunsaturated fatty acids to the drug may reduce the risk of dose-related side effects in patients treated with OCA.
2025-07-22
Biochemical Journal (published)
doi.org
Pharmaco-nutraceutical improvement of the response to obeticholic acid with omega-3 polyunsaturated fatty acids
Audrey-Anne Lavoie
Ariane Thérien
Anisia Silva
Emanuel Paré
Anna Ciešlak
William Gagnon
Clémence Desjardins
Mélanie Verreault
Jocelyn Trottier
Marie-Claude Vohl
Jean-Philippe Drouin-Chartier
Jacques Corbeil
Alexandre Caron
Olivier Barbier
Obeticholic acid (OCA) is the second-line therapy for primary biliary cholangitis. While efficient in promoting bile acid (BA) detoxificatio… (see more)n and limiting liver fibrosis, its clinical use is restricted by severe dose-dependent side effects. We tested the hypothesis that adding n-3 polyunsaturated fatty acids (PUFAs), eicosapentaenoic (EPA) and docosahexaenoic (DHA) acids to OCA may improve the therapeutic effect of the low drug dosage. Several liver cell lines were exposed to vehicle, low or high OCA dose (1–20 μM) in the presence or absence of EPA/DHA for 24 h. To induce ER stress, apoptosis, and fibrosis, HepG2 cells were exposed to a 400 μM BA mixture or to 2 ng/ml transforming growth factor-β (TGF-β). For inflammation analyses, THP-1 cells were activated with 100 ng/ml lipopolysaccharides (LPS). The impact of OCA+EPA/DHA was assessed using transcriptomic (qRT-PCR), proteomic (ELISA, caspase-3), and metabolomic (LC-MS/MS) approaches. The addition of EPA/DHA reinforced the ability of low OCA dose to down-regulate the expression of genes involved in BA synthesis (CYP7A1 and CYP8B1) and uptake (NTCP) and to up-regulate the expression of MRP2 and 3 genes. EPA/DHA also enhanced the anti-inflammatory response of the drug by reducing the expression of the LPS-induced cytokines: tumor necrosis factor α (TNFα), interleukin (IL)-6, IL-1β, and monocyte chemoattractant protein-1 in THP-1 macrophages. OCA+EPA/DHA decreased the expression of BIP, CHOP, and COL1A1 genes and the caspase-3 activity. EPA+DHA potentiate the response to low OCA doses on BA toxicity and provide additional benefits on ER stress, apoptosis, inflammation, and fibrosis. These observations support the idea that adding n-3 PUFAs to the drug may reduce the risk of dose-related side effects in patients treated with OCA.
2025-07-22
Biochemical Journal (published)
doi.org