Yufan Guo

Jiacheng Lin, Zhongruo Wang, Kun Qian et al.

Supervised Fine-Tuning (SFT) on domain-specific datasets is a common approach to adapt Large Language Models (LLMs) to specialized tasks but is often believed to degrade their general capabilities. In this work, we revisit this trade-off and present both empirical and theoretical insights. First, we show that SFT does not always hurt: using a smaller learning rate can substantially mitigate general performance degradation while preserving comparable target-domain performance. We then provide a theoretical analysis that explains these phenomena and further motivates a new method, Token-Adaptive Loss Reweighting (TALR). Building on this, and recognizing that smaller learning rates alone do not fully eliminate general-performance degradation in all cases, we evaluate a range of strategies for reducing general capability loss, including L2 regularization, LoRA, model averaging, FLOW, and our proposed TALR. Experimental results demonstrate that while no method completely eliminates the trade-off, TALR consistently outperforms these baselines in balancing domain-specific gains and general capabilities. Finally, we distill our findings into practical guidelines for adapting LLMs to new domains: (i) using a small learning rate to achieve a favorable trade-off, and (ii) when a stronger balance is further desired, adopt TALR as an effective strategy.

Zhe Liu, Yufan Guo, Jalal Mahmud

Although deep neural networks have been widely employed and proven effective in sentiment analysis tasks, it remains challenging for model developers to assess their models for erroneous predictions that might exist prior to deployment. Once deployed, emergent errors can be hard to identify in prediction run-time and impossible to trace back to their sources. To address such gaps, in this paper we propose an error detection framework for sentiment analysis based on explainable features. We perform global-level feature validation with human-in-the-loop assessment, followed by an integration of global and local-level feature contribution analysis. Experimental results show that, given limited human-in-the-loop intervention, our method is able to identify erroneous model predictions on unseen data with high precision.

Mehdi Moradi, Ali Madani, Yaniv Gur et al.

Medical image analysis practitioners have embraced big data methodologies. This has created a need for large annotated datasets. The source of big data is typically large image collections and clinical reports recorded for these images. In many cases, however, building algorithms aimed at segmentation and detection of disease requires a training dataset with markings of the areas of interest on the image that match with the described anomalies. This process of annotation is expensive and needs the involvement of clinicians. In this work we propose two separate deep neural network architectures for automatic marking of a region of interest (ROI) on the image best representing a finding location, given a textual report or a set of keywords. One architecture consists of LSTM and CNN components and is trained end to end with images, matching text, and markings of ROIs for those images. The output layer estimates the coordinates of the vertices of a polygonal region. The second architecture uses a network pre-trained on a large dataset of the same image types for learning feature representations of the findings of interest. We show that for a variety of findings from chest X-ray images, both proposed architectures learn to estimate the ROI, as validated by clinical annotations. There is a clear advantage obtained from the architecture with pre-trained imaging network. The centroids of the ROIs marked by this network were on average at a distance equivalent to 5.1% of the image width from the centroids of the ground truth ROIs.

Yiding Lu, Yufan Guo, Anna Korhonen

Background: In silico drug-target interaction (DTI) prediction plays an integral role in drug repositioning: the discovery of new uses for existing drugs. One popular method of drug repositioning is network-based DTI prediction, which uses complex network theory to predict DTIs from a drug-target network. Currently, most network-based DTI prediction is based on machine learning methods such as Restricted Boltzmann Machines (RBM) or Support Vector Machines (SVM). These methods require additional information about the characteristics of drugs, targets and DTIs, such as chemical structure, genome sequence, binding types, causes of interactions, etc., and do not perform satisfactorily when such information is unavailable. We propose a new, alternative method for DTI prediction that makes use of only network topology information attempting to solve this problem. Results: We compare our method for DTI prediction against the well-known RBM approach. We show that when applied to the MATADOR database, our approach based on node neighborhoods yield higher precision for high-ranking predictions than RBM when no information regarding DTI types is available. Conclusion: This demonstrates that approaches purely based on network topology provide a more suitable approach to DTI prediction in the many real-life situations where little or no prior knowledge is available about the characteristics of drugs, targets, or their interactions.