Alzheimer's disease (AD) detection from spontaneous speech provides a noninvasive and scalable alternative to traditional diagnostics but remains challenged by limited labeled data in target languages. This paper presents a modular multimodal method for robust AD detection in label-scarce languages without language-specific supervision. The approach integrates three complementary feature streams-acoustic, paralinguistic, and semantic-through dedicated modules: ARFU for cognitively salient feature compression, DIFA for directed intra-modal refinement, and MASA for task-adaptive modality integration. On the ADReSS-M benchmark, our method achieves 91.30% accuracy and 90.91% F1 on the Greek test set, surpassing the previous best by 4.61% and maintaining strong robustness in zero- and few-shot settings. These results confirm the effectiveness of structured fusion and language-agnostic modeling for reliable AD detection in label-scarce language scenarios.

We present DuST-Net, a dual-stream temporal network for early Alzheimer's disease screening from spontaneous speech. The model explicitly decouples and jointly models microlevel intra-sentential dynamics and macro-level inter-sentential discourse flow via two parallel streams, which together ingest a four-component feature set: paralinguistic cues, global discourse descriptors, and sentence-level acoustic and linguistic features. Finally, DuST-Net is trained and evaluated on the PROCESS dataset. The results show that our model outperforms all baselines and significantly improves the Macro-F1 Score. Additional analyses demonstrate that differentiating between temporal scales and modalities is crucial for achieving balanced and robust AD screening from speech.

Knowledge tracing assesses students’ mastery and predicts future performance based on historical learning data. Traditional methods primarily rely on predefined static associations between concepts and exercises, which struggle to capture potential causal relationships and dynamic learning patterns, leading to reduced prediction accuracy and limited interpretability. To address these issues, this paper proposes a novel dynamic causal inference framework that integrates Gumbel-Softmax sampling with uncertainty estimation, transforming discrete causal relationships into differentiable continuous weights, and quantifying model uncertainty to enhance robustness against noisy data and improve interpretability. Additionally, inspired by item response theory, the model dynamically adjusts students’ latent states by modeling the interaction between student ability and exercises difficulty. Experimental results on three widely-used benchmarks demonstrate that this method achieves state-of-the-art (SOTA) performance in prediction accuracy while also generating interpretable causal relationship weights that provide insights into knowledge acquisition patterns.

Polyp segmentation plays a critical role in the diagnosis of colorectal cancer but remains challenging due to the diverse morphologies and indistinct boundaries of polyps. These challenges are particularly prominent in small and irregularly shaped polyps, where precise edge detail capture is essential. While the Segment Anything Model 2 (SAM2) has demonstrated remarkable progress in medical image segmentation through its strong pre-training capabilities, its reliance on global feature extraction strategies limits its ability to model local details and fine-grained features, particularly for complex edges. To address these shortcomings, we propose MSA-SAM2Net, an enhanced framework that integrates a Large Kernel Size Attention (LKSA) module and an Edge-Boosting Module (EBM) to improve feature extraction and edge detail perception. Specifically, LKSA employs multi-scale and gating mechanisms to capture both global context and local details, producing richer attention maps. Meanwhile, EBM enhances segmentation accuracy by focusing on edge regions. Additionally, we introduce a Large-Kernel Grouped Attention (LGA) to further refine feature selection and fusion. Comprehensive experiments on five publicly available and challenging polyp segmentation datasets demonstrate the effectiveness of MSA-SAM2Net, achieving state-of-the-art performance on all five datasets. The source code will be available at: https://github.com/coderMelon0216/MSA-SAM2Net

Label quality critically impacts the performance of federated learning.However, client data often contains varying degrees of label noise.To address this, we propose FedDMC, a framework featuring a dual-model dynamic interaction mechanism to mitigate error accumulation in high-noise scenarios.FedDMC integrates multi-stage noise correction, and an adaptive regularization strategy to improve robustness against complex noise distributions.We evaluate FedDMC on benchmark datasets under diverse heterogeneous noise settings and compare it with state-of-the-art methods, demonstrating its superior performance.Furthermore, we validate its applicability in realworld medical imaging tasks using ADNI-MRI and ADNI-PET datasets.Experimental results confirm that FedDMC consistently outperforms existing approaches under complex heterogeneous label noise.

Detecting cognitive biases in large language models (LLMs) is a fascinating task that aims to probe the existing cognitive biases within these models. Current methods for detecting cognitive biases in language models generally suffer from incomplete detection capabilities and a restricted range of detectable bias types. To address this issue, we introduced the ‘MindScope’ dataset, which distinctively integrates static and dynamic elements. The static component comprises 5,170 open-ended questions spanning 72 cognitive bias categories. The dynamic component leverages a rule-based, multi-agent communication framework to facilitate the generation of multi-round dialogues. This framework is flexible and readily adaptable for various psychological experiments involving LLMs. In addition, we introduce a multi-agent detection method applicable to a wide range of detection tasks, which integrates Retrieval-Augmented Generation (RAG), competitive debate, and a reinforcement learning-based decision module. Demonstrating substantial effectiveness, this method has shown to improve detection accuracy by as much as 35.10% compared to GPT-4. Codes and appendix are available at https://github.com/2279072142/MindScope.

Few-shot text classification has attracted great interest in both academia and industry due to the lack of labeled data in many fields. Different from general text classification (e.g., topic classification), few-shot sentiment classification is more challenging because the semantic distances among the classes are more subtle. For instance, the semantic distances between the sentiment labels in a positive or negative polarity (e.g., "love" and "joy", "remorse" and "sadness") are close, while the distances are large for the sentiment labels in two opposite polarities (e.g., "love" and "sadness"). To address this problem, we propose a Soft Contrastive learning-based Prompt (SCP) model for few-shot sentiment analysis. First, we design a sentiment-aware chain of thought prompt module to guide the model to predict the sentiment from coarse grain to fine grain via a series of intermediate reasoning steps. Then, we propose a soft contrastive learning algorithm to take the correlation of the labels into account. A series of experiments on several sentiment analysis datasets show the great advantages of SCP by comparing it with SOTA baselines (e.g., ChatGPT).

The field of image generation models has seen substantial progress, characterized by a proliferation of diverse generative models and their associated outputs. However, there currently exists a deficiency in methodologies that can concurrently and effectively evaluate both the intrinsic quality of generated images and the alignment between image features and textual prompts. To address these challenges, we propose a novel Framework for evaluating Visual elegance and Sentiment resonance (FVS). The FVS incorporates a novel image aesthetic assessment model, specifically trained to assess the visual attractiveness of the generated images. Additionally, it evaluates the sentiment and aesthetic consistency between textual prompt and the generated image. Experimental results verify that the evaluations from our framework align more closely with human preferences. Moreover, we apply our framework to filter and construct a higher-quality training set of generated images. This curated dataset is then exploited to adapt the generative model, resulting in enhanced generation quality.

Few-shot class-incremental learning (FSCIL) aims to learn novel concepts using limited examples without forgetting. However, most studies focus on solving the catastrophic forgetting problem of FSCIL, while neglecting the poor robustness of these algorithms. For example, introducing just a 10% FGSM attack can result in a decrease of more than 30% in accuracy. To tackle this challenge, we propose the Vital Importance Playback for Robust FSCIL (VIP-FSCIL) method. The effectiveness of this approach stems from its consideration of sample importance to construct a replay exemplar set, as well as the reduction of overfitting risks to adversarial examples through knowledge distillation. Our approach thus enhances the robustness of FSCIL models while preserving their generalization. The experimental results indicate that VIP-FSCIL is able to improve the accuracy of the model in perturbed scenarios by 5-9% on CIFAR100 and CUB200 datasets, compared to other methods.

In a real-world setting, machine needs to continuously recognize new categories without forgetting. However, the number of new categories may be small. For some difficult categories, even humans cannot recognize only based on few-shot examples. To address the above issues, an innovative uncertainty-aware few-shot class incremental learning method (UACL) is proposed, which allows the model to continuously recognize new classes with few-shot examples and identify the classes it cannot recognize currently. Besides, in order to imitate the cognitive way of human beings and improve the continuous representation ability, we propose a pseudo-incremental task construction mechanism based on uncertainty estimation, where the machine learn to recognize from simple to difficult. Further, a large-scale pre-training model is used as an expert system to guide the model to recognize difficult classes. We evaluate our method on three popular benchmark datasets, showing that UACL is state-of-the-art.

Building a good feature space is essential for the metric-based few-shot algorithms to recognize a novel class with only a few samples. The feature space is often built by Convolutional Neural Networks (CNNs). However, CNNs primarily focus on local information with the limited receptive field, and the global information generated by distant pixels is not well used. Meanwhile, having a global understanding of the current task and focusing on distinct regions of the same sample for different queries are important for the few-shot classification. To tackle these problems, we propose the Cross Non-Local Neural Network (CNL) for capturing the long-range dependency of the samples and the current task. CNL extracts the task-specific and context-aware features dynamically by strengthening the features of the sample at a position via aggregating information from all positions of itself and the current task. To reduce losing important information, we maximize the mutual information between the original and refined features as a constraint. Moreover, we add a task-specific scaling to deal with multi-scale and task-specific features extracted by CNL. We conduct extensive experiments for validating our proposed algorithm, which achieves new state-of-the-art performances on two public benchmarks.

Few-shot learning can recognize a novel category based on only a few samples because it learns to learn from a lot of labeled samples during the training process. When data is insufficient, the performance is affected. And it is expensive to obtain a large-scale finegrained dataset with annotation. In this paper, we adopt domain- specific knowledge to fill the gap of insufficient annotated data. We propose a cross-modal knowledge distillation (CMKD) framework to do fine-grained one-shot classification and propose the Spatial Relation Loss (SRL) to transfer cross-modal information, which can tackle the semantic gap between multimodal features. The teacher network distills the spatial relationship of the samples as a soft target for training a unimodal student network. Notably, the student network makes predictions only based on a few samples without any external knowledge in the application. This model-agnostic framework will be well adapted to other few-shot models. Extensive experimental results on benchmarks demonstrate that CMKD can make full use of cross-modal knowledge in image and text few-shot classification. CKMD improves the performances of the student networks significantly, even if it is a state-of-the-art student network.

In this article, we propose an online and unsupervised anomaly detection algorithm for streaming data using an array of sliding windows and the probability density-based descriptors (PDDs) (based on these windows). This algorithm mainly consists of three steps: 1) we use a main sliding window over streaming data and segment this window into an array of nonoverlapping subwindows; 2) we propose the PDDs with dimension reduction, based on the kernel density estimation, to estimate the probability density of data in each subwindow; and 3) we design the distance-based anomaly detection rule to determine whether the current observation is anomalous. The experimental results and performances are presented based on the Numenta anomaly benchmark. Compared with the anomaly detection algorithm using the hierarchical temporal memory proposed by Numenta (which outperforms a wide range of other anomaly detection algorithms), our algorithm can perform better in many cases, that is, with higher detection rates and earlier detection for contextual anomalies and concept drifts.

The small inter-class variance and the large intra-class variance make the few-shot and fine-grained image classification more difficult because the machine cannot obtain enough information from only a few images. The external knowledge contains more semantics and can support the model to extract important features, while most of existing few-shot learning algorithms only focus on leveraging the visual features from images, little attention has been paid to the cross-modal external knowledge. In this paper, we propose a knowledge-based fine-grained classification mechanism for few-shot learning, which can overcome the difficulty of only obtaining limited and discriminative features from unimodal samples. We extract the visual features and the knowledge features from textual descriptions and a domain-specific knowledge graph at global and local levels to build the semantic space. To tackle the gap between multimodal features, we propose a mirror framework, named Mirror Mapping Network (MMN), to map the multimodal features into the same semantic space with two directions. Extensive experimental results show that our method outperforms the state-of-the-art.

In this paper, we introduce a brand new intersection setting which combines a reconfigured stop line and variable message signs unit, along with vehicle-to-infrastructure and infrastructure-to-infrastructure communications. A buffer waiting zone resulting from the new stop line contributes to compensating the starting time from a standstill and shortening the intersection crossing time. An improved traffic light control algorithm is adapted for the proposed stop line, then both human-driven vehicles and connected and automated vehicles can distinctly determine their strategies only from the current phase without considering the remaining phase time. Compared with pre-timed and actuated traffic signal systems, simulations validate the effectiveness of our method.

This paper considers a target-defense problem of one defender and one adversary (or called intruder), in which the defender tries to approach a desired interception-position between the intruder and the target to intercept the intruder and expel it from the target. The defender adopts this interception and expelling strategy since the defender is assumed to not capture or destroy the intruder. An expelling-decay exponent is introduced to characterize the expelling-decay rate on the intruder. The system is nonlinear, with distinct physical meaning and rich kinematics. The main concern in this paper is the analysis of the kinematic properties. First, two motion patterns of the system are characterized with respect to different values of the system parameters. Then, the stability and the transition condition of the two motion patterns are provided. Finally, the optimal interception of the defender is provided, which interestingly coincides with the transition condition for the two motion patterns. The interpretations for the physical meaning of the optimal interception are also provided.

For most attribute reduction in Pawlak rough set model (PRS), monotonicity is a basic property for the quantitative measure of an attribute set. Based on the monotonicity, a series of attribute reductions in Pawlak rough set model such as positive-region-preserved reductions and condition entropy-preserved reductions are defined and the corresponding heuristic algorithms are proposed in previous rough sets research. However, some quantitative measures of attribute set may be non-monotonic in probabilistic rough set model such as decision-theoretic rough set (DTRS), and the non-monotonic definition of the attribute reduction should be reinvestigated and the heuristic algorithm should be reconsidered. In this paper, the monotonicity of the positive region in PRS and DTRS are comparatively discussed. Theoretic analysis shows that the positive region in DTRS model may be expanded with the decrease of the attributes, which is essentially different from that in PRS model. Hereby, a new non-monotonic attribute reduction is presented for the DTRS model in this paper, and a heuristic algorithm for searching the newly defined attribute reduction is proposed, in which the positive region is allowed to be expanded instead of remaining unchanged in the process of attribute reduction. Experimental analysis is included to validate the theoretic analysis and quantify the effectiveness of the proposed attribute reduction algorithm.

With increasing demands on software functions, software systems become more and more complex. This complexity is one of the most pervasive factors affecting software development productivity. Assessing the impact of software complexity on development productivity helps to provide effective strategies for development process and project management. Previous research literatures have suggested that development productivity declines exponentially with software complexity. Borrowing insights from cognitive learning psychology and behavior theory, the relationship between software complexity and development productivity was reexamined in this paper. This research identified that the relationship partially showed a U-shaped as well as an inverted U-shaped curvilinear tendency. Furthermore, the range of complexity level that is beneficial for productivity has been presented, in which, the lower bound denotes the minimum degree of complexity at which personnel can be motivated, while the upper bound shows the maximum extent of complexity that staff can endure. Based on our findings, some guidelines for improving personnel management of software industry have also been given.