This paper proposes Attention-Seeker, an unsupervised keyphrase extraction method that leverages self-attention maps from a Large Language Model to estimate the importance of candidate phrases. Our approach identifies specific components – such as layers, heads, and attention vectors – where the model pays significant attention to the key topics of the text. The attention weights provided by these components are then used to score the candidate phrases. Unlike previous models that require manual tuning of parameters (e.g., selection of heads, prompts, hyperparameters), Attention-Seeker dynamically adapts to the input text without any manual adjustments, enhancing its practical applicability. We evaluate Attention-Seeker on four publicly available datasets: Inspec, SemEval2010, SemEval2017, and Krapivin. Our results demonstrate that, even without parameter tuning, Attention-Seeker outperforms most baseline models, achieving state-of-the-art performance on three out of four datasets, particularly excelling in extracting keyphrases from long documents.

Traditional knowledge graph (KG) approach often rely on static textbook content and overlook the dynamic, collaborative interactions in collaborative problem-solving (CPS). This study introduced a three-step integrated KG approach designed to support CPS in STEM education and examined the effective KG features that influence CPS learning outcomes. KGs were generated by combining learning materials and student dialogue data. Two types of features, graph structural features and knowledge activation features, were identified to classify knowledge nodes and analyze how students activated knowledge during CPS. Clustering analysis revealed three types of knowledge nodes: Peripheral Nodes, Core Nodes, and Degree Hubs. Furthermore, key features such as depth, branch, and activated paths showed positive correlations with group discussion performance and CPS skills but had limited influence on test scores. These findings highlight the potential of integrated KGs to support both individual and group learning in STEM education.

Time series modeling serves as the cornerstone of real-world applications, such as weather forecasting and transportation management. Recently, Mamba has become a promising model that combines near-linear computational complexity with high prediction accuracy in time series modeling, while facing challenges such as insufficient modeling of nonlinear dependencies in attention and restricted receptive fields caused by convolutions. To overcome these limitations, this paper introduces an innovative framework, Attention Mamba, featuring a novel Adaptive Pooling block that accelerates attention computation and incorporates global information, effectively overcoming the constraints of limited receptive fields. Furthermore, Attention Mamba integrates a bidirectional Mamba block, efficiently capturing long-short features and transforming inputs into the Value representations for attention mechanisms. Extensive experiments conducted on diverse datasets underscore the effectiveness of Attention Mamba in extracting nonlinear dependencies and enhancing receptive fields, establishing superior performance among leading counterparts. Our codes will be available on GitHub.

In the domain of artificial neural networks, the comprehensive understanding and optimization of neuron dynamics are crucial. The Dendritic Neuron Model (DNM), noted for its distinct architecture and data processing capabilities, exemplifies this. However, the sophistication of the DNM leads to complexities in hyperparameter tuning. Notably, this complexity manifests in the way parameter changes can alter the dimensions of the solution space, a prime example of the Metameric Variable-length Problem. In this study, we have innovatively integrated chaotic maps into the gene expression mechanisms of Evolutionary Algorithms, enabling the incorporation of all DNM hyperparameters into a singular algorithmic framework. This integration allows for iterative adjustments within a variable-dimensional solution space, representing an evolving neuron model that streamlines the tuning process. Our approach, tested against benchmark datasets from the UCI Machine Learning Repository, demonstrates significant improvements in the DNM's performance, highlighting the effectiveness of incorporating biological chaos phenomena into neural network optimization.

The use of digital textbooks in education provides valuable data on student reading behavior that can help educators refine their course materials and instructional design for future iterations. Previous studies have explored methods for extracting important evidence from this data, but they require manual intervention. By automating these methods, this paper introduces an end-to-end system capable of extracting evidence from e-book data and providing recommendations for slides' content review based on this evidence. Our system incorporates information about reading preferences into the evidence-extraction process and implements Large Language Models (LLMs) for automatic interpretation. Six teachers evaluated our proposed system indicating a promising level of effectiveness, while also highlighting areas for future improvement to ensure a successful classroom implementation. These include considerations for improving the actionability of recommendations, improving the identification of content that needs refinement, and improving the performance of LLMs.

Binocular vision serves as the foundation for stereo vision, providing humans with the ability to perceive the three-dimensional information of their surroundings. However, although some cortical neurons are reported with 3D information selectivity, the interconnections between binocular input and cortical neuron firing have yet to be fully established, impeding progress in our understanding of stereo information perception. In this work, we aim to address this issue by examining the causality between retinal input and selective neuron firing. We propose a general mechanism of stereo orientation and motion direction detection based solely on binocular disparity input, which is the difference in visual information between the two eyes. Our results suggest that this disparity-based mechanism is robust and can effectively complete stereo orientation and motion direction detection. Our proposed general perceiving mechanism has the potential to contribute to the resolution of the complex problem of binocular information processing and computation. Further research into this area may help to deepen our understanding of stereo vision and provide insights into the underlying neural mechanisms.

A novel coronavirus discovered in late 2019 (COVID-19) quickly spread into a global epidemic and, thankfully, was brought under control by 2022. Because of the virus's unknown mutations and the vaccine's waning potency, forecasting is still essential for resurgence prevention and medical resource management. Computational efficiency and long-term accuracy are two bottlenecks for national-level forecasting. This study develops a novel multivariate time series forecasting model, the densely connected highly flexible dendritic neuron model (DFDNM) to predict daily and weekly positive COVID-19 cases. DFDNM's high flexibility mechanism improves its capacity to deal with nonlinear challenges. The dense introduction of shortcut connections alleviates the vanishing and exploding gradient problems, encourages feature reuse, and improves feature extraction. To deal with the rapidly growing parameters, an improved variation of the adaptive moment estimation (AdamW) algorithm is employed as the learning algorithm for the DFDNM because of its strong optimization ability. The experimental results and statistical analysis conducted across three Japanese prefectures confirm the efficacy and feasibility of the DFDNM while outperforming various state-of-the-art machine learning models. To the best of our knowledge, the proposed DFDNM is the first to restructure the dendritic neuron model's neural architecture, demonstrating promising use in multivariate time series prediction. Because of its optimal performance, the DFDNM may serve as an important reference for national and regional government decision-makers aiming to optimize pandemic prevention and medical resource management. We also verify that DFDMN is efficiently applicable not only to COVID-19 transmission prediction, but also to more general multivariate prediction tasks. It leads us to believe that it might be applied as a promising prediction model in other fields.

As a recurrent spiking neural network, liquid state machine (LSM) has attracted more and more attention in neuromorphic computing due to its biological plausibility, computation power, and hardware implementation. However, the neural architecture of LSM, such as hidden neuron number, synaptic density, percentage connectivity, and connection state, has significant impact on its model performance. Manually defining a neural architecture will be ineffective and laborious in most cases. Therefore, based on a state-of-the-art differential evolution algorithm, an evolutionary neural architecture design methodology is proposed to automatically build suitable model topologies for LSM in this study, without any prior knowledge. The effectiveness of the proposed method has been validated on commonly-used image classification tasks.