Efficient and reliable data collection is essential for agricultural Internet of Things (IoT) systems, where timely sensing supports precision farming. Unmanned aerial vehicle (UAV)-assisted collection provides flexible and low cost coverage. However, in wide and remote fields, UAV-assisted collection faces two coupled challenges. First, agricultural field operations and field-state changes, such as irrigation, rainfall, harvesting, canopy occlusion, machinery movement, and fluctuations in solar exposure, can jointly and unevenly perturb traffic, communication, and energy processes, leading to bursty arrivals, degraded air-to-ground (A2G) links, and variations in harvested energy. Second, under single-link communication, the UAV can directly access only one node in each slot, making frequent global state refresh difficult and causing stale beliefs to accumulate rapidly after abrupt changes. These effects lead to biased scheduling, unnecessary maneuvering, packet loss, and excess energy consumption. To address this problem, this paper proposes a three stage framework for joint node scheduling and UAV trajectory control. It first refreshes selected node states to correct stale beliefs and then encodes refreshed and inferred states through a reliability-aware Transformer. Finally, it performs conservative hybrid-action control to coordinate discrete scheduling and continuous motion while mitigating value overestimation under partial observability and non-stationarity. Simulations under point-shift, cluster-shift, and global-shift scenarios show that the proposed method consistently reduces packet loss, improves energy efficiency, and achieves faster recovery than representative reinforcement learning (RL) and heuristic baselines.

Artificial Intelligence Generated Content (AIGC) has emerged as a mainstream research direction with the development of Large Language Models (LLMs). The hallucination of LLMs, however, always interweaves the generated content with outdated or fabricated information, making it hard to be fully trusted and severely hindering LLMs form being widely applied in real-life scenarios. To address this problem, Retrieval Augmented Generation (RAG) has been proposed, which incorporates external knowledge to assist LLMs with content generation and significantly alleviates the hallucination problem. Nonetheless, the vanilla RAG uses similarity as the sole criterion for selecting external knowledge, neglecting the problem of internal inconsistency within the knowledge itself, which may distract LLMs from focusing on the most important information during the content generation process and, therefore, has a negative impact on the generated content's reliability. In this paper, we propose a novel metric, Entropy-based Internal Consistency (EIC), to measure the internal consistency of the external knowledge which is then integrated with similarity to mutually determine the knowledge's importance. Experimental results demonstrate that the proposed metric can provide a more fine-grained signal for external knowledge selection, thereby enhancing the reliability of generated content.

Anomalous sound detection (ASD) is crucial in industrial applications due to its non-invasive and real-time capabilities. However, existing ASD methods often rely on autoencoders, which require machine-specific tuning, or large pre-trained models with high computational costs. Additionally, many self-supervised approaches depend on extensive meta-information, increasing deployment complexity. To address these limitations, we propose a lightweight, metadata-free ASD framework that generalizes across different machine types without requiring complex hyperparameter tuning. Our approach extracts high-dimensional features from Log-Mel spectrograms using MobileNetV2, then refines feature representations through relational learning with a Graph Neural Network-based SAGE-GAT model. Unlike conventional methods that treat machine types independently, our approach leverages cross-category feature propagation through local neighbor relationships, capturing discriminative information from nearby samples. Furthermore, an MLP optimized with ArcFace loss enhances feature structuring, while anomaly detection is performed using K-means clustering. Experiments on the DCASE 2024 Task 2 dataset validate the effectiveness of our approach, demonstrating its robustness, efficiency, and suitability for real-world industrial deployment.

In the field of very large-scale integration (VLSI) chip design, chip floorplanning plays a crucial role as it directly influences key optimization objectives such as placement wirelength. This, in turn, affects signal delay, power efficiency, routability, and overall cost. However, traditional reinforcement learning (RL) methods for chip floorplanning often oversimplify this complex and dynamic task. They tend to overlook the cascading effects of module placements and fail to fully comprehend the intricate interdependencies that are vital for making informed decisions. To address these challenges, we introduce an innovative approach that combines hypergraph graph convolutional networks (HyperGCNs) with deep recurrent Q-networks (DRQNs). This integration allows us to capture the nuanced dynamics and interconnected aspects of chip design more effectively. We enhance the traditional Markov decision process (MDP) model by incorporating a state characterization layer based on long short-term memory (LSTM) technology. Initially, HyperGCN efficiently encodes netlist information, simplifying complex graph structures into lower dimensional vectors, thereby enhancing knowledge processing. Subsequently, we treat the chip as an agent and apply the DRQN algorithm to optimize the module layout. DRQN’s LSTM utilizes a recurrent layer structure to grasp dependencies between modules, combined with the deep Q-network (DQN)’s optimization capabilities, enabling us to navigate the complexities of floorplanning. Our approach improves the state representation by encompassing a broader understanding of interconnected module characteristics. This allows our agents to make decisions that take into account the collective impact of module adjustments, rather than viewing each change in isolation. This comprehensive state representation, which includes diverse features and their evolving relationships, significantly enhances the decision-making capabilities of our agents. Our extensive experiments demonstrate that our method outperforms traditional heuristic-based and other learning-based floorplanning techniques. To the best of our knowledge, this is the first application of LSTM with DQN in circuit design, representing a significant advancement in the field.

The multi-agent characteristics of integrated energy systems are becoming increasingly prominent, rendering the energy management problems more intricate. Additionally, the distributed agents are faced with challenges from the communication layer, such as packet losses and privacy disclosure issues. Therefore, a privacy-preserving neurodynamic-based optimization strategy considering packet losses is proposed in this article. A privacy-preserving communication model is constructed based on differential privacy (DP). Unlike traditional DP methods, the privacy preservation mechanism employed in this article can achieve a quantified high level of privacy while ensuring the solution accuracy of the optimization algorithm. Moreover, communication packet loss models based on both the two-state Markov process and the Bernoulli process are established. The proposed fully distributed scheme only requires communication between adjacent agents. The efficiency of the neurodynamic-based approach, the high-level privacy preservation, and the robustness against communication packet loss are demonstrated through several case studies.

In the realm of very large-scale integration (VLSI) chip design, chip floorplanning is essential, directly impacting key optimization objectives like placement wirelength, which in turn affects signal delay, power efficiency, routability, and overall cost. Traditional reinforcement learning (RL) methods for chip floorplanning often oversimplify the complex, dynamic nature of this task, typically overlooking the cascading effects of module placements and failing to fully grasp the intricate interdependencies crucial for informed decision-making. To address these challenges, we introduce an innovative approach that fuses hypergraph graph convolutional networks (HyperGCN) with deep Transformer Q-networks (DTQN). This integration captures the nuanced dynamics and interconnected aspects of chip design more effectively. We enhance the traditional Markov decision process (MDP) model with a state characterization layer based on Transformer technology. Initially, HyperGCN effectively encodes netlist information, simplifying complex graph structures into lower-dimensional vectors, thus enhancing knowledge processing. Subsequently, viewing the chip as an agent, we apply the DTQN algorithm to optimize the module layout. DTQN's Transformer encoder utilizes a multi-head self-attention (MSA) mechanism to grasp long-range dependencies between modules, coupled with DQN's optimization capabilities, to navigate the complexities of floorplanning. Our approach enhances the state representation by incorporating a broader understanding of the interconnected module characteristics, which allows agents to make decisions that account for the collective impact of module adjustments, rather than viewing each change in isolation. This comprehensive state representation, encompassing diverse features and their evolving relationships, significantly enhances the agent's decision-making capabilities. Our extensive experiments demonstrate that our method outperforms traditional heuristic-based and other learning-based floorplanning techniques. To our knowledge, this is the first application of the Transformer in circuit design, marking a significant advancement in the field.

The finite inverted beta mixture model (IBMM) has been proven to be efficient in modeling positive vectors. Under the traditional variational inference framework, the critical challenge in Bayesian estimation of the IBMM is that the computational cost of performing inference with large datasets is prohibitively expensive, which often limits the use of Bayesian approaches to small datasets. An efficient alternative provided by the recently proposed stochastic variational inference (SVI) framework allows for efficient inference on large datasets. Nevertheless, when using the SVI framework to address the non-Gaussian statistical models, the evidence lower bound (ELBO) cannot be explicitly calculated due to the intractable moment computation. Therefore, the algorithm under the SVI framework cannot directly use stochastic optimization to optimize the ELBO, and an analytically tractable solution cannot be derived. To address this problem, we propose an extended version of the SVI framework with more flexibility, namely, the extended SVI (ESVI) framework. This framework can be used in many non-Gaussian statistical models. First, some approximation strategies are applied to further lower the ELBO to avoid intractable moment calculations. Then, stochastic optimization with noisy natural gradients is used to optimize the lower bound. The excellent performance and effectiveness of the proposed method are verified in real data evaluation.

With the rapid development of the internet worldwide, network security issues are becoming increasingly prominent. Network intrusion detection systems (NIDSs) play a vital role in ensuring computer network security due to their ability to identify potential network threats. Despite considerable research efforts, deploying NIDSs on resource-constrained devices has been challenging. To reduce the imposed computational cost and model storage requirements, in this paper, we propose a novel lightweight NIDS model. In this model, patterns of normal and malicious actions are learned via a finite Dirichlet mixture model (DMM) in the context of the extended stochastic variational inference (ESVI) framework. With the proposed method, both the parameter estimation and model selection processes can be simultaneously addressed in a unified Bayesian framework. A great number of experiments conducted on three publicly available datasets demonstrate that the proposed model not only achieves comparable classification performance to that of detection models based on several well-studied finite mixture modeling, traditional machine learning (ML) and promising deep learning (DL) algorithms but also significantly reduces the required training and detection time. Extensive experimental results validate that the proposed model is a feasible and efficient lightweight intrusion detection model.

Currently, speech emotion recognition models still could not show satisfactory performance due to the complexity of emotions. In most of the previous studies, there is a common problem that some of the particular emotions are severely misclassified. In this article, we propose a novel framework integrating cascaded attention network and adversarial joint loss strategy for speech emotion recognition, aiming at discriminating the confusions by emphasizing more on the emotions which are difficult to be correctly classified. First, we extract log-Mels, deltas and delta-deltas of log-Mels as 3D features to effectively reduce the interference of external factors. Next, we introduce a cascaded attention network to extract effective emotional features, where spatiotemporal attention selectively locates the targeted emotional regions from the input features. In these targeted regions, the self attention with head fusion captures the long-distance dependence of temporal features. Finally, an adversarial joint loss strategy is proposed to distinguish the emotional embeddings with high similarity by the generated hard triplets in an adversarial fashion. To evaluate our proposed method, experiments are performed with the IEMOCAP, CASIA, and EMODB corpora. The experimental results demonstrate that our proposed method significantly outperforms the state-of-the-art approaches on all datasets.

High temperature or temperature nonuniformity has been considered as one of the most challenging problems in three-dimensional integrated circuits (3-D ICs). There have been many studies on the thermal issues in 3-D ICs floorplanning. However, most handcrafted heuristic algorithms require long iteration cycles, resulting in inefficient thermal management with no guarantee of good performance. Meanwhile, as modern integrated circuit design becomes increasingly complex, current floorplannings suffer from the “curse of dimensionality” and cannot optimize large-scale cases. Therefore, a novel thermal-aware fixed-outline 3-D IC floorplanning is proposed in this article. Since the temperature in a 3-D IC is mainly determined by a power distribution across tiers, this article proposes a deep${k}$-means clustering algorithm to cluster the modules and find a better cross-tier power distribution. This is an unsupervised machine-learning algorithm that can successfully cluster high-dimensional sample data. Then, in the global distribution (GD) stage, we not only consider the power consumption between the groups but also apply the number and area of through silicon vias (TSVs) to the clustering and analytical method in the multilevel framework. Inspired by the fact that finding the optimal TSV positions for multiple nets in the TSV assignment (TA) stage is essentially a way to maximize the benefits of nets in a competitive environment, game theory is applied to further reduce temperature and wirelength in this stage. We prove that it is an ordinal potential game, and then converge to the Nash equilibrium by the best response strategy. Experimental results demonstrate that the proposed method shows good performance in reducing the temperature of 3-D ICs, and its running time is quite fast.

High temperature and temperature nonuniformity pose significant challenges in 3-D integrated circuits (3-D ICs). Numerous studies have explored thermal issues in 3-D IC floorplanning. However, most existing handcrafted heuristic algorithms suffer from long iteration cycles, resulting in inefficient thermal management and no guarantee of optimal performance. In addition, with the increasing complexity of modern integrated circuit design, current floorplanning techniques encounter the “curse of dimensionality” and struggle to optimize large-scale cases. To address these challenges, this article proposes a novel end-to-end learning-based approach for thermal-aware fixed-outline 3-D IC floorplanning. In the tier assignment stage, we utilize a deep${k}$-means clustering algorithm to allocate modules to different tiers, aiming to achieve an improved cross-tier power distribution. In the global distribution (GD) stage, we formulate the floorplanning problem as a Markov decision process (MDP). By combining graph convolutional networks (GCNs) with a multiagent deep reinforcement learning (MADRL) algorithm, we optimize the positions of modules and through-silicon vias (TSVs), while incorporating an attention mechanism in the centralized critic to enhance cooperation among agents. Finally, in the TSV assignment (TA) stage, we refine the TSV positions using the MADRL algorithm, further reducing wirelength and temperature in 3-D ICs. Experimental results demonstrate that our proposed approach outperforms state-of-the-art heuristic-based 3-D IC floorplanner in terms of wirelength and temperature optimization.

Three-dimensional integrated circuit (3-D IC) technology has emerged as a solution to address the limitations of 2-D ICs. One critical component of 3-D ICs is the through-silicon via (TSV), which plays a crucial role in connecting signal nets and dissipating heat. The assignment of TSVs significantly impacts the wirelength and temperature of 3-D ICs. Although several studies have focused on the TSV assignment problem, most existing heuristic algorithms suffer from long iteration cycles and lack performance guarantees. Meanwhile, these methods struggle to optimize large-scale cases due to the curse of dimensionality inherent in complex IC designs. Recently, learning-based algorithms, particularly reinforcement learning (RL), have demonstrated success in solving various combinatorial optimization problems by leveraging past experiences. In this article, we propose attention-TSV assignment (ATT-TA), a novel TSV assignment method based on a multiagent deep RL (MADRL) algorithm. The TSV assignment problem is formulated as a Markov decision process (MDP), where each TSV layer serves as an independent agent. To enhance collaboration, we incorporate an attention mechanism that models and utilizes teammate strategies within a cooperative multiagent system. By applying multiagent systems, we mitigate the curse of dimensionality, while the attention mechanism enables adaptive modeling of joint strategies among agents. Notably, this is the first attempt to solve the TSV assignment problem using a MADRL algorithm. The experimental results demonstrate that compared with the state-of-the-art heuristic-based TSV assignment methods, the proposed ATT-TA approach consistently outperforms these methods in terms of wirelength and temperature optimization.

Bayesian estimation of parameters in the Dirichlet mixture process of the Beta-Liouville distribution (i.e., the infinite Beta-Liouville mixture model) has recently gained considerable attention due to its modeling capability for proportional data. However, applying the conventional variational inference (VI) framework cannot derive an analytically tractable solution since the variational objective function cannot be explicitly calculated. In this paper, we adopt the recently proposed extended VI framework to derive the closed-form solution by further lower bounding the original variational objective function in the VI framework. This method is capable of simultaneously determining the model's complexity and estimating the model's parameters. Moreover, due to the nature of Bayesian nonparametric approaches, it can also avoid the problems of underfitting and overfitting. Extensive experiments were conducted on both synthetic and real data, generated from two real-world challenging applications, namely, object detection and text categorization, and its superior performance and effectiveness of the proposed method have been demonstrated.