We use cookies to analyze the browsing and usage of our website and to personalize your experience. You can disable these technologies at any time, but this may limit certain functionalities of the site. Read our Privacy Policy for more information.
Setting cookies
You can enable and disable the types of cookies you wish to accept. However certain choices you make could affect the services offered on our sites (e.g. suggestions, personalised ads, etc.).
Essential cookies
These cookies are necessary for the operation of the site and cannot be deactivated. (Still active)
Analytics cookies
Do you accept the use of cookies to measure the audience of our sites?
Multimedia Player
Do you accept the use of cookies to display and allow you to watch the video content hosted by our partners (YouTube, etc.)?
Publications
Multi-Objective Risk Assessment Framework for Exploration Planning Using Terrain and Traversability Analysis
Personalized medicine based on medical images, including predicting future individualized clinical disease progression and treatment respons… (see more)e, would have an enormous impact on healthcare and drug development, particularly for diseases (e.g. multiple sclerosis (MS)) with long term, complex, heterogeneous evolutions and no cure. In this work, we present the first stochastic causal temporal framework to model the continuous temporal evolution of disease progression via Neural Stochastic Differential Equations (NSDE). The proposed causal inference model takes as input the patient's high dimensional images (MRI) and tabular data, and predicts both factual and counterfactual progression trajectories on different treatments in latent space. The NSDE permits the estimation of high-confidence personalized trajectories and treatment effects. Extensive experiments were performed on a large, multi-centre, proprietary dataset of patient 3D MRI and clinical data acquired during several randomized clinical trials for MS treatments. Our results present the first successful uncertainty-based causal Deep Learning (DL) model to: (a) accurately predict future patient MS disability evolution (e.g. EDSS) and treatment effects leveraging baseline MRI, and (b) permit the discovery of subgroups of patients for which the model has high confidence in their response to treatment even in clinical trials which did not reach their clinical endpoints.
Efficiently quantifying predictive uncertainty in medical images remains a challenge. While Bayesian neural networks (BNN) offer predictive … (see more)uncertainty, they require substantial computational resources to train. Although Bayesian approximations such as ensembles have shown promise, they still suffer from high training and inference costs. Existing approaches mainly address the costs of BNN inference post-training, with little focus on improving training efficiency and reducing parameter complexity. This study introduces a training procedure for a sparse (partial) Bayesian network. Our method selectively assigns a subset of parameters as Bayesian by assessing their deterministic saliency through gradient sensitivity analysis. The resulting network combines deterministic and Bayesian parameters, exploiting the advantages of both representations to achieve high task-specific performance and minimize predictive uncertainty. Demonstrated on multi-label ChestMNIST for classification and ISIC, LIDC-IDRI for segmentation, our approach achieves competitive performance and predictive uncertainty estimation by reducing Bayesian parameters by over 95\%, significantly reducing computational expenses compared to fully Bayesian and ensemble methods.
MiRGraph: A hybrid deep learning approach to identify microRNA-target interactions by integrating heterogeneous regulatory network and genomic sequences
MicroRNAs (miRNAs) mediates gene expression regulation by targeting specific messenger RNAs (mRNAs) in the cytoplasm. They can function as b… (see more)oth tumor suppressors and oncogenes depending on the specific miRNA and its target genes. Detecting miRNA-target interactions (MTIs) is critical for unraveling the complex mechanisms of gene regulation and promising towards RNA therapy for cancer. There is currently a lack of MTIs prediction methods that simultaneously perform feature learning from heterogeneous gene regulatory network (GRN) and genomic sequences. To improve the prediction performance of MTIs, we present a novel transformer-based multiview feature learning method – MiRGraph, which consists of two main modules for learning the sequence-based and GRN-based feature embedding. For the former, we utilize the mature miRNA sequences and the complete 3’UTR sequence of the target mRNAs to encode sequence features using a hybrid transformer and convolutional neural network (CNN) (TransCNN) architecture. For the latter, we utilize a heterogeneous graph transformer (HGT) module to extract the relational and structural information from the GRN consisting of miRNA-miRNA, gene-gene and miRNA-target interactions. The TransCNN and HGT modules can be learned end-to-end to predict experimentally validated MTIs from MiRTarBase. MiRGraph outperforms existing methods in not only recapitulating the true MTIs but also in predicting strength of the MTIs based on the in-vitro measurements of miRNA transfections. In a case study on breast cancer, we identified plausible target genes of an oncomir.
MiRGraph: A transformer-based feature learning approach to identify microRNA-target interactions by integrating heterogeneous graph network and sequence information
MicroRNAs (miRNAs) play a crucial role in the regulation of gene expression by targeting specific mRNAs. They can function as both tumor sup… (see more)pressors and oncogenes depending on the specific miRNA and its target genes. Detecting miRNA-target interactions (MTIs) is critical for unraveling the complex mechanisms of gene regulation and identifying therapeutic targets and diagnostic markers. There is currently a lack of MTIs prediction method that simultaneously performs feature learning on heterogeneous graph network and sequence information. To improve the prediction performance of MTIs, we present a novel transformer-based multi-view feature learning method, named MiRGraph. It consists of two main modules for learning the sequence and heterogeneous graph network, respectively. For learning the sequence-based feaature embedding, we utilize the mature miRNA sequence and the complete 3’UTR sequence of the target mRNAs to encode sequence features. Specifically, a transformer-based CNN (TransCNN) module is designed for miRNAs and genes respectively to extract their personalized sequence features. For learning the network-based feature embedding, we utilize a heterogeneous graph transformer (HGT) module to extract the relational and structural information in a heterogeneous graph consisting of miRNA-miRNA, gene-gene and miRNA-target interactions. We learn the TransCNN and HGT modules end-to-end by utilizing a feedforward network, which takes the combined embedded features of the miRNA-gene pair to predict MTIs. Comparisons with other existing MTIs prediction methods illustrates the superiority of MiRGraph under standard criteria. In a case study on breast cancer, we identified plausible target genes of an oncomir hsa-MiR-122-5p and plausible miRNAs that regulate the oncogene BRCA1.
We study the depth of grade-school math (GSM) problem-solving capabilities of LLMs. To this end, we evaluate their performance on pairs of e… (see more)xisting math word problems together so that the answer to the second problem depends on correctly answering the first problem. Our findings reveal a significant reasoning gap in most LLMs, that is performance difference between solving the compositional pairs and solving each question independently. This gap is more pronounced in smaller, more cost-efficient, and math-specialized models. Moreover, instruction-tuning recipes and code generation have varying effects across LLM sizes, while finetuning on GSM can lead to task overfitting. Our analysis indicates that large reasoning gaps are not because of test-set leakage, but due to distraction from additional context and poor second-hop reasoning. Overall, LLMs exhibit systematic differences in their reasoning abilities, despite what their performance on standard benchmarks indicates.
Autonomous robots navigating in off-road terrain like forests open new opportunities for automation. While off-road navigation has been stud… (see more)ied, existing work often relies on clearly delineated pathways. We present a method allowing for long-range planning, exploration and low-level control in unknown off-trail forest terrain, using vision and GPS only. We represent outdoor terrain with a topological map, which is a set of panoramic snapshots connected with edges containing traversability information. A novel traversability analysis method is demonstrated, predicting the existence of a safe path towards a target in an image. Navigating between nodes is done using goal-conditioned behavior cloning, leveraging the power of a pretrained vision transformer. An exploration planner is presented, efficiently covering an unknown off-road area with unknown traversability using a frontiers-based approach. The approach is successfully deployed to autonomously explore two 400 meters squared forest sites unseen during training, in difficult conditions for navigation.
Large language models (LLMs) are increasingly applied to complex reasoning tasks that require executing several complex steps before receivi… (see more)ng any reward. Properly assigning credit to these steps is essential for enhancing model performance. Proximal Policy Optimization (PPO), a state-of-the-art reinforcement learning (RL) algorithm used for LLM finetuning, employs value networks to tackle credit assignment. However, value networks face challenges in predicting the expected cumulative rewards accurately in complex reasoning tasks, often leading to high-variance updates and suboptimal performance. In this work, we systematically evaluate the efficacy of value networks and reveal their significant shortcomings in reasoning-heavy LLM tasks, showing that they barely outperform a random baseline when comparing alternative steps. To address this, we propose VinePPO, a straightforward approach that leverages the flexibility of language environments to compute unbiased Monte Carlo-based estimates, bypassing the need for large value networks. Our method consistently outperforms PPO and other RL-free baselines across MATH and GSM8K datasets with fewer gradient updates (up to 9x), less wall-clock time (up to 3.0x). These results emphasize the importance of accurate credit assignment in RL finetuning of LLM and demonstrate VinePPO's potential as a superior alternative.
