With the rapid advancement of deep learning, drug target interaction (DTI) prediction has seen substantial performance enhancements. However, existing methodologies face a critical, yet unaddressed challenge, i.e., the Modality Reliability Gap. Such a gap arises from the unpredictable variance in the informativeness and reliability of 1D sequence versus 3D structural data across different drug-target pairs, critically limiting model robustness and domain generalization capabilities. To overcome it, we introduce DrugCMF, a novel Drug-Target interaction prediction method via Confidence-aware Multimodal Fusion framework designed specifically to bridge the Modality Reliability Gap. Specifically, the DrugCMF employs a four-stage approach: (1) it extracts rich features by utilizing four pre-trained models to obtain token-level embeddings from both 1D sequences and 3D structures. (2) it preserves modality informativeness by independently learning interaction patterns within each modality through a Token-level Interaction module. (3) it explicitly quantifies the reliability gap by employing a novel confidence estimation mechanism to dynamically learn weights for each modality. (4) it bridges the gap by using these confidence scores to guide a learnable cross-modal fusion module, adaptively fusing information from the most trustworthy source. By methodically addressing the Modality Reliability Gap, DrugCMF significantly outperforms SOTA methods.

Test-time few-shot object detection (FSOD) represents an innovative approach for identifying novel categories using a limited number of support examples, obviating the need for model fine-tuning. Despite advancements, existing FSOD methods, including our prior work, continue to grapple with challenges posed by domain/category shift and limited data availability. Building upon our previous research on test-time FSOD, this article proposes a novel dynamic prototype fusion network (PFN) to overcome these limitations. To mitigate the impact of the distribution shift, a dynamic prototype refinement method is introduced that updates prototypes from supporting images in an adaptive manner. Further, limited samples are mitigated through exhaustive exploitation of information within support images. Specifically, we design a dual-level multiscale information integration approach that effectively fuses information across different network layers and image scales, enhancing the model's discriminating capabilities. Additionally, a mask-based preprocessing technique harnesses segmentation labels on support samples, effectively suppressing the adverse impact of background noise on model accuracy. Notably, to align with the constraints of test-time scenarios, model parameters remain fixed during the configuration step, with only prototypes being updated each time users input novel supporting samples. As a result, our method achieves superior performance over existing state-of-the-art FSOD methods on multiple benchmarks, demonstrating remarkable potential in the realm of FSOD. The code is available at https://github.com/CatfishW/TIDEV2.

Current Multi-Modal Large language Models (MMLMs) primarily rely on instance-level feature statistics for cross-modal alignment. However, they commonly suffer three inherent limitations including vulnerability to outlier perturbations, neglect of inter-feature covariance structures, and local optimum trapping. These limitations stem from a critical oversight—existing approaches disregard the global statistical structure of multi-modal data, treating cross-modal alignment as isolated feature-level alignment rather than systematic distribution-level alignment. To address these issues, this paper proposes Layer-wise Covariance Alignment (LCA), which first leverages distribution-level alignment for cross-modal alignment. The effectiveness of LCA is validated through the use of parameter-efficient Low-Rank Adaptation (LoRA) on CLIP architectures. Experimental validation across eight benchmarks demonstrates state-of-the-art performance, confirming the critical role of distribution-level alignment in overcoming sample-level optimization constraints for cross-modal learning.

Few-shot object detection (FSOD) aims to extract semantic knowledge from limited object instances of novel categories within a target domain. Recent advances in FSOD focus on fine-tuning the base model based on a few objects via meta-learning or data augmentation. Despite their success, the majority of them are grounded with parametric readjustment to generalize on novel objects, which face considerable challenges in Industry 5.0, such as 1) a certain amount of fine-tuning time is required and 2) the parameters of the constructed model being unavailable due to the privilege protection, making the fine-tuning fail. Such constraints naturally limit its application in scenarios with real-time configuration requirements or within black-box settings. To tackle the challenges mentioned above, we formalize a novel FSOD task, referred to as test-time few-shot detection (TIDE), where the model is un-tuned in the configuration procedure. To that end, we introduce an asymmetric architecture for learning a support-instance-guided dynamic category classifier. Further, a cross-attention module and a multiscale resizer are provided to enhance the model performance. Experimental results on multiple FSOD platforms reveal that the proposed TIDE significantly outperforms existing contemporary methods. The implementation codes are available at https://github.com/deku-0621/TIDE.

Unsupervised domain adaptation (UDA) aims to transfer knowledge from a well-labeled source domain to a related and unlabeled target domain with identical label space. The main workhorse in UDA is domain alignment and has proven successful. However, it is practically difficult to find an appropriate source domain with identical label space. A more practical scenario is partial domain adaptation (PDA) where the source label space subsumes the target one. Unfortunately, due to the non-identity between label spaces, it is extremely hard to obtain an ideal alignment, conversely, easier resulting in mode collapse and negative transfer. These motivate us to find a relatively simpler alternative to solve PDA. To achieve this, we first explore a theoretical analysis, which says that the target risk is bounded by both model smoothness and between-domain discrepancy. Then, we instantiate the model smoothness as an intra-domain structure preserving (IDSP) while giving up possibly riskier domain alignment. To our best knowledge, this is the first naive attempt for PDA without alignment. Finally, our empirical results on benchmarks demonstrate that IDSP is not only superior to the PDA SOTAs (e.g., ∼ +10% on Cl → Rw and ∼ +8% on Ar → Rw), but also complementary to domain alignment in the standard UDA.

The functional connectivity network (FCN) has been used to achieve several remarkable advancements in the diagnosis of neuro-degenerative disorders. Therefore, it is imperative to accurately estimate biologically meaningful FCNs. Several efforts have been dedicated to this purpose by encoding biological priors. However, owing to the high complexity of the human brain, the estimation of an 'ideal' FCN remains an open problem. To the best of our knowledge, almost all existing studies lack the integration of domain expert knowledge, which limits their performance. In this study, we focused on incorporating domain expert knowledge into the FCN estimation from a modularity perspective. To achieve this, we presented a human-guided modular representation (MR) FCN estimation framework. Specifically, we designed an adversarial low-rank constraint to describe the module structure of FCNs under the guidance of domain expert knowledge (i.e., a predefined participant index). The chronic tinnitus (TIN) identification task based on the estimated FCNs was conducted to examine the proposed MR methods. Remarkably, MR significantly outperformed the baseline and state-of-the-art(SOTA) methods, achieving an accuracy of 92.11%. Moreover, post-hoc analysis revealed that the FCNs estimated by the proposed MR could highlight more biologically meaningful connections, which is beneficial for exploring the underlying mechanisms of TIN and diagnosing early TIN.

Mild cognitive impairment (MCI) is an intermediate stage of brain cognitive decline, associated with increasing risk of developing Alzheimer's disease (AD). It is believed that early treatment of MCI could slow down the progression of AD, and functional brain network (FBN) could provide potential imaging biomarkers for MCI diagnosis and response to treatment. However, there are still some challenges to estimate a “good” FBN, particularly due to the poor quality and limited quantity of functional magnetic resonance imaging (fMRI) data from the target domain (i.e., MCI study). Inspired by the idea of transfer learning, we attempt to transfer information in high-quality data from source domain (e.g., human connectome project in this paper) into the target domain towards a better FBN estimation, and propose a novel method, namely NERTL (Network Estimation via Regularized Transfer Learning). Specifically, we first construct a high-quality network “template” based on the source data, and then use the template to guide or constrain the target of FBN estimation by a weighted l1-norm regularizer. Finally, we conduct experiments to identify subjects with MCI from normal controls (NCs) based on the estimated FBNs. Despite its simplicity, our proposed method is more effective than the baseline methods in modeling discriminative FBNs, as demonstrated by the superior MCI classification accuracy of 82.4% and the area under curve (AUC) of 0.910.

Functional brain network (FBN) is becoming an increasingly important measurement for exploring cerebral mechanisms and mining informative biomarkers that assist diagnosis of some neurodegenerative disorders. Despite its effectiveness to discover valuable hidden patterns in the human brain, the estimated FBNs are often heavily influenced by the quality of the observed data (e.g., blood oxygen level dependent signal series). In practice, a preprocessing pipeline is usually employed for improving data quality. With this in mind, some data points (volumes or time course in the time series) are still not clean enough, due to artifacts including spurious resting-state processes (head movement, mind-wandering). Therefore, not all volumes in the fMRI time series can contribute to the subsequent FBN estimation. To address this issue, we propose a novel FBN estimation method by introducing a latent variable as an indicator of the data quality, and develop an alternating optimization algorithm for jointly scrubbing the data and estimating FBN simultaneously. To further illustrate the effectiveness of the proposed method, we conduct experiments on two public datasets to identify subjects with mild cognitive impairment from normal controls based on the estimated FBNs, and achieve improved accuracies than the baseline methods.