As a pivotal variant of multi-label classification, hierarchical text classification (HTC) faces unique challenges due to its intricate taxonomic hierarchy. Recent state-of-the-art approaches improve performance by considering both global hierarchy covering all labels and local hierarchy indicating substructure of sample-specific ground-truth labels. However, they often over-condense hierarchical information into one or several tokens, which may cause the loss of useful knowledge. Accordingly, we propose a dual classifier model with global and local hierarchies (DCGL). It adopts prompt tuning-based BERT as the backbone, where global hierarchy is integrated into the soft prompt template. And this resulting classifier branch is termed global pipeline. To mitigate information loss caused by hierarchy condensation, we introduce a parallel local hierarchy-aware classifier pipeline. This local pipeline acquires label-level classification features through text propagation on the label hierarchy and aligns these features with oracle label representations of local hierarchy via graph contrastive learning, which serve as a novel strategy for local hierarchy incorporation. Thereby, DCGL obtains more granular and targeted features and captures local hierarchy information such as label co-occurrence and local structure. Moreover, since global and local pipelines capture distinct yet complementary information, we further apply mutual knowledge distillation to bridge the gap between their output logits and facilitate mutual learning. And to better control the distillation degree, we design a dynamic temperature negatively correlated with label confidence. Comprehensive experiments demonstrate that our DCGL outperforms several representative HTC methods.

Understanding how students think—not merely what they know—remains the central challenge of knowledge tracing. Current approaches model knowledge as a flat collection of concepts and are fundamentally unable to capture the hierarchical reasoning processes that govern human problem-solving. We introduce SPARK-KT ( S tepwise P rerequisite- A ware R easoning for K nowledge T racing), a framework that reconceptualizes knowledge tracing through the lens of cognitive dependency structures. SPARK-KT leverages Large Language Models to automatically decompose each question into a Directed Acyclic Graph of reasoning steps, where nodes represent atomic cognitive operations, and edges encode prerequisite constraints. A novel dependency-aware mastery propagation mechanism ensures that a student’s competence on any step is bounded by their weakest prerequisite—mirroring the pedagogical reality that complex skills cannot be deployed without foundational mastery. Through a dual-stream architecture that disentangles conceptual knowledge from procedural structure, SPARK-KT improves predictive accuracy while providing fine-grained interpretability: it identifies not just whether a student will succeed, but which specific reasoning step constitutes their cognitive bottleneck. Experiments across three diverse educational domains demonstrate consistent improvements over sixteen state-of-the-art baselines, with particularly strong gains on multi-step reasoning problems. Beyond prediction, SPARK-KT enables actionable diagnostics—revealing latent potential masked by prerequisite gaps and informing precisely targeted interventions.

Partial Maximum Satisfiability (PMS) and Weighted Partial Maximum Satisfiability (WPMS) generalize Maximum Satisfiability (MaxSAT), with broad real-world applications. Recent advances in Stochastic Local Search (SLS) algorithms for solving (W)PMS have mainly focused on designing clause weighting schemes. However, existing methods often fail to adequately distinguish between PMS and WPMS, typically employing uniform update strategies for clause weights and overlooking critical structural differences between the two problem types. In this work, we present a novel clause weighting scheme that, for the first time, updates the clause weights of PMS and WPMS instances according to distinct conditions. This scheme also introduces a new initialization method, which better accommodates the unique characteristics of both instance types. Furthermore, we propose a decimation method that prioritizes satisfying unit and hard clauses, effectively complementing our proposed clause weighting scheme. Building on these methods, we develop a new SLS solver for (W)PMS named DeepDist. Experimental results on benchmarks from the anytime tracks of recent MaxSAT Evaluations show that DeepDist outperforms state-of-the-art SLS solvers. Notably, a hybrid solver combining DeepDist with TT-Open-WBO-Inc surpasses the performance of the MaxSAT Evaluation 2024 winners, SPB-MaxSAT-c-Band and SPB-MaxSAT-c-FPS, highlighting the effectiveness of our approach. The code is available at https://github.com/jmhmaxsat/DeepDist

The prediction of breast cancer survival rates holds important clinical significance for treatment decision-making by attending physicians and treatment selection by patients. According to the characteristics of the problem, this paper proposes a multilayer perceptron (MLP) based stacking ensemble learning approach, Stacking-MLP , and apply it to the analysis of survival rates among breast cancer patients within 1-year, 3-year, 5-year, and 10-year periods. Stacking-MLP consists of two stages: firstly, it trains a series of primary perceptron models through feature random undersampling to mitigate the negative impact of inter-feature correlations; then, it employs a fully connected neural network for ensemble learning (Stacking ensemble learning) to adaptively extract complementary information from the learning results of the primary perceptrons. The Surveillance, Epidemiology, and End Results(SEER) is an authoritative cancer epidemiology database managed by the National Cancer Institute in the United States. In this study, we constructed a SEER dataset consisting of 19 variables for breast cancer survival prediction studies. We evaluated the performance of the proposed Stacking-MLP model on this dataset, and the experimental results show an improvement of approximately 10% in accuracy compared to similar studies. The experimental code and the constructed dataset used in this study are publicly available at https://github.com/GaoHaokai222/Stacking-MLP.

Histopathological image analysis is the gold standard for breast cancer diagnosis. In recent years, the pathological image recognition technology based on deep learning has attracted great attention. However, a current issue with existing methods is the lack of effective extraction of overall contextual information and key details from the images. To this end, we proposed an efficient pathological image recognition model CSAM-ResNet based on channel attention mechanism, spatial attention mechanism, residual learning principle and multi-scale fusion method. The spatial and channel attention mechanism (CSAM) combines the advantages of capturing long-distance feature dependencies in Transformer and local feature extraction in CNN. Therefore, CSAM-ResNet can extract the global context information in the image more comprehensively, while strengthening the sensitivity to the key information. To evaluate the effectiveness of the proposed method in this paper, we experimentaly compared the CSAM-ResNet with SE-ResNet and SwinTransformer on the authoritative breast cancer pathological image classification datasets PatchCamelyon (Pcam) and BreakHis. It is shown that CBAM-ResNet improves classification accuracy by 0.3% and 0.4% in the Pcam and BreakHis datasets, respectively. The code is available on site: https://github.com/luckyrz0/CSAM-ResNet.