While Large Reasoning Models (LRMs) exhibit remarkable capabilities in complex tasks, they often suffer from excessive redundancy in their chain-of-thought reasoning. This significantly reduces inference efficiency and increases computational costs. We identify that LRM redundancy is not uniformly homogeneous but can be taxonomized according to whether it is destructive to the final answer: destructive redundancy (e.g., logical drift, hallucination amplification) versus non-destructive redundancy (e.g., repetition, over-elaboration). Moreover, LRM's redundant and concise responses exhibit a significant distinction in their hidden layer representation spaces. Based on these insights, we propose CATS (Category-Aware Token-level Steering), a training-free and lightweight method to reduce the redundancy phenomenon. CATS decomposes redundancy into six semantically interpretable characteristic dimensions. By flexibly weighting and combining the differential vectors corresponding to these dimensions, CATS synthesizes a composite intervention vector, enabling zero-parameter intervention in the hidden layers. Experiments across three LRM models and five mathematical reasoning datasets demonstrate that CATS reduces reasoning length by an average of 25% while maintaining or even slightly improving task accuracy. CATS offers a pluggable, training-free, and lightweight solution, making it particularly beneficial for users in low-resource environments.

Modeling user-item interactions in a dynamic manner bring new insight to the representation learning for recom-mender systems. Distinct from static graph-based approaches that model the whole user-item interaction graph, dynamic graph-based approaches model both the structural and temporal information from a sequence of snapshot graphs. Despite effectiveness, we argue that existing approaches do not explicitly address the temporal sparsity issue, which degrades the representation learning performance for inactive users and items. Therefore, we propose a new Dynamic Graph-based Collaborative Filtering(DynGCF) framework. In particular, it utilizes the vanilla interaction graph with the co-occurrence graph(co-graph) to jointly explores 1-hop collaborative and 2-hop implicit similarity for dynamic representation learning. Moreover, to further alleviate temporal sparsity, we explore representative(active) users and items via graph pooling and design an activity-guided gating(AGate) layer to augment inactive users and items. At last, we further stack a temporal aggregator layer to obtain the final representation. We conduct extensive experiments on four real-world benchmark datasets to demonstrate the significant performance gains for DynGCF over several state-of-the-art methods. Further analyses also show the necessity of alleviating temporal sparsity for improving recommendation performance.

Recently, sequence features have been extensively studied to improve the performance of recommender systems. However, advanced sequential recommendation methods that rely only on item IDs still face the challenge of modeling fine-grained user preference from interactive data. Furthermore, context-aware sequential recommendations have the hardness of modeling the relationship between items and items, items and users. Both of these two methods ignore the effect of categories on users' next click tendency and the interactive learning between categories and items. In this paper, we propose a method named Contextual Collaborative Graph Attention Network (CCGAT) to model the sequence. Methodologically, user behavior sequences are constructed as graph-structured data, and we apply two similar graph self-attention networks to model the item transitions and the category click probability. CCGAT takes advantage of the fact that users tend to click on the same or similar categories under specific purposes, and provides a simple but effective way to train two networks collaboratively. Extensive experiments on five real-world datasets show that our model outperforms state-of-the-art methods, and demonstrate the validity of modeling both contextual information and graph features.

Session-based recommendation aims to predict the next item that users will interact based solely on anonymous sessions. In real-life scenarios, the user's preferences are usually various, and distinguishing different preferences in the session is important. However previous studies focus mostly on the transition modeling between items, ignoring the mining of various user preferences. In this paper, we propose a Hierarchical Leaping Network (HLN) to explicitly model the users' multiple preferences by grouping items that share some relationships. We first design a Leap Recurrent Unit (LRU) which is capable of skipping preference-unrelated items and accepting knowledge of previously learned preferences. Then we introduce a Preference Manager (PM) to manage those learned preferences and produce an aggregated preference representation each time LRU reruns. The final output of PM which contains multiple preferences of the user is used to make recommendations. Experiments on two benchmark datasets demonstrate the effectiveness of HLN. Furthermore, the visualization of explicitly learned subsequences also confirms our idea.

Recent years has witnessed growing interest in hyperspectral image (HSI) processing. In practice, however, HSIs always suffer from huge data size and mass of redundant information, which hinder their application in many cases. HSI compression is a straightforward way of relieving these problems. However, most of the conventional image encoding algorithms mainly focus on the spatial dimensions, and they need not consider the redundancy in the spectral dimension. In this paper, we propose a novel HSI compression and reconstruction algorithm via patch-based low-rank tensor decomposition (PLTD). Instead of processing the HSI separately by spectral channel or by pixel, we represent each local patch of the HSI as a third-order tensor. Then, the similar tensor patches are grouped by clustering to form a fourth-order tensor per cluster. Since the grouped tensor is assumed to be redundant, each cluster can be approximately decomposed to a coefficient tensor and three dictionary matrices, which leads to a low-rank tensor representation of both the spatial and spectral modes. The reconstructed HSI can then be simply obtained by the product of the coefficient tensor and dictionary matrices per cluster. In this way, the proposed PLTD algorithm simultaneously removes the redundancy in both the spatial and spectral domains in a unified framework. The extensive experimental results on various public HSI datasets demonstrate that the proposed method outperforms the traditional image compression approaches and other tensor-based methods.

This paper concentrates on reducing the cost of I/O and improving algorithm efficiency by designing and realizing a spatial data access system based on main memory database (MMDB). We use Redis to manage vector spatial data in main memory which is faster than hard disk or disk-based database in respect of data access. We design a management strategy of spatial data, internal and external data exchange strategy and data backup mechanism. Based on MMDB, we design the model of vector spatial data to describe its form, organization and access methods. The experimental results show that our data access strategy can cut down the time of I/O and improve the algorithm efficiency effectively.

Since the hyperspectral images (HSI) could provide much more useful discriminative information that cannot be obtained by the conventional imaging techniques, the hyper-spectral imaging technology was widely used in remote sensing area and recently used in many other aspects, such as the biological images recognition. However, most of the time, the size of hyperspectral data is so large that to process these data is both time-consuming and space-consuming. In this paper, a multiway tensor projection (MTP) algorithm is proposed as an extension to the conventional PCA for hyperspectral data compression and reconstruction. Technologically speaking, MTP carries out a tensor data compression in all the modes simultaneously to seek a projection matrix along each order to make sure that the projected core tensor can preserve most of the information present in the original tensor. Since the MTP algorithm uses the arbitrary order tensor as the input, it can preserve the structure information not only among the rows and columns but also among the spectral channels as much as possible and without vectorization. Numerous experiments on hyperspectral biological databases show that the MTP algorithm has better compression performance than PCA in many aspects.