Neural Processing Letters最新文献

Neural networks as differential equation solvers are a good choice of numerical technique because of their fast solutions and their nature in tackling some classical problems which traditional numerical solvers faced. In this article, we look at the famous gradient descent optimization technique, which trains the network by updating parameters which minimizes the loss function. We look at the theoretical part of gradient descent to understand why the network works great for some terms of the loss function and not so much for other terms. The loss function considered here is built in such a way that it incorporates the differential equation as well as the derivative of the differential equation. The fully connected feed-forward network is designed in such a way that, without training at boundary points, it automatically satisfies the boundary conditions. The neural tangent kernel for gradient enhanced physics informed neural networks is examined in this work, and we demonstrate how it may be used to generate a closed-form expression for the kernel function. We also provide numerical experiments demonstrating the effectiveness of the new approach for several two point boundary value problems. Our results suggest that the neural tangent kernel based approach can significantly improve the computational accuracy of the gradient enhanced physics informed neural network while reducing the computational cost of training these models.

Multimodal sentiment analysis is a downstream branch task of sentiment analysis with high attention at present. Previous work in multimodal sentiment analysis have focused on the representation and fusion of modalities, capturing the underlying semantic relationships between modalities by considering contextual information. While this approach is feasible for simple contextual comments, more complex comments require the integration of external knowledge to obtain more accurate sentiment information. However, incorporating external knowledge into sentiment analysis to enhance information complementarity has not been thoroughly investigated. To address this, we propose a multichannel cross-modal feedback interaction model that incorporates the knowledge graph into multimodal sentiment analysis. Our proposed model consists of two main components: the cross-modal feedback recurrent interaction module and the external knowledge module for capturing latent information. The cross-modal interaction employs a self-feedback mechanism during network training, extracting feature representations of each modality and using these representations to mask sensory inputs, allowing the model to perform feedback-based feature masking. The external knowledge graph captures potential semantic information representations in the textual data through knowledge graph embedding. Finally, a global feature fusion module is employed for multichannel multimodal information integration. On two publicly available datasets, our method demonstrates good performance in terms of accuracy and F1 scores, compared to state-of-the-art models and several baselines.

This paper proposes a quantized model-free adaptive iterative learning control (MFAILC) algorithm to solve the bipartite containment tracking problem of unknown nonlinear multi-agent systems, where the interactions between agents include cooperation and antagonistic interactions. To design the controller, the agent’s dynamics is transformed into the linear data model based on the dynamic linearization method, and then a quantized MFAILC algorithm is established based on the quantized values of the relative output measurements. The designed controller only depends on the input and output data of the agent. We prove that under the quantized MFAILC algorithm, the multi-agent systems can achieve the bipartite containment, that is, the output trajectories of followers converge to the convex hull formed by the leaders’ trajectories and the leaders’ symmetric trajectories. Finally, we provide simulations to illustrate the effectiveness of our theoretical results.

The development of high-stakes decision-making neural agents that interact with complex environments, such as video games, is an important aspect of AI research with numerous potential applications. Reinforcement learning combined with deep learning architectures (DRL) has shown remarkable success in various genres of games. The performance of DRL is heavily dependent upon the neural networks resides within them. Although these algorithms perform well in offline testing but the performance deteriorates in noisy and sub-optimal conditions, creating safety and security issues. To address these, we propose a hybrid deep learning architecture that combines a traditional convolutional neural network with worm brain-inspired neural circuit policies. This allows the agent to learn key coherent features from the environment and interpret its dynamics. The obtained DRL agent was not only able to achieve an optimal policy quickly, but it was also the most noise-resilient with the highest success rate. Our research indicates that only 20 control neurons (12 inter-neurons and 8 command neurons) are sufficient to achieve competitive results. We implemented and analyzed the agent in the popular video game Doom, demonstrating its effectiveness in practical applications.

Existing pre-trained models have yielded promising results in terms of computational time reduction. However, these models only focus on pruning simple sentences or less salient words, while neglecting the treatment of relatively complex sentences. It is frequently these sentences that cause the loss of model accuracy. This shows that the adaptation of the existing models is one-sided. To address this issue, in this paper, we propose a sample-adaptive training and inference model. Specifically, complex samples are extracted from the training datasets and a dedicated data augmentation module is trained to extract global and local semantic information of complex samples. During inference, simple samples can exit the model via the Sample Adaptive Exit Mechanism, Normal samples pass through the whole backbone model before inference, while complex samples are processed by the Characteristic Enhancement Module after passing through the backbone model. In this way, all samples are processed adaptively. Our extensive experiments on classification tasks datasets in the field of Natural Language Processing demonstrate that our method enhances model accuracy and reduces model inference time for multiple datasets. Moreover, our method is transferable and can be applied to multiple pre-trained models.

This paper investigates the topology identification and synchronization in finite time for fractional singularly perturbed complex networks (FSPCNs). Firstly, a convergence principle is developed for continuously differential functions. Secondly, a dynamic event-triggered mechanism (DETM) is designed to achieve the network synchronization, and a topology observer is developed to identify the network topology. Thirdly, under the designed DETM, by constructing a Lyapunov functional and applying the inequality analysis technique, the topology identification and synchronization condition in finite time is established in the forms of the matrix inequality. In addition, it is proved that the Zeno behavior can be effectively excluded. Finally, the effectiveness of the main results is verified by an application example.

Video anomaly event detection is crucial for analyzing surveillance videos. Existing methods have limitations: frame-level detection fails to remove background interference, and object-level methods overlook object-environment interaction. To address these issues, this paper proposes a novel video anomaly event detection algorithm based on a dual-channel autoencoder with key region feature enhancement. The goal is to preserve valuable information in the global context while focusing on regions with a high anomaly occurrence. Firstly, a key region extraction network is proposed to perform foreground segmentation on video frames, eliminating background redundancy. Secondly, a dual-channel autoencoder is designed to enhance the features of key regions, enabling the model to extract more representative features. Finally, channel attention modules are inserted between each deconvolution layer of the decoder to enhance the model’s perception and discrimination of valuable information. Compared to existing methods, our approach accurately locates and focuses on regions with a high anomaly occurrence, improving the accuracy of anomaly event detection. Extensive experiments are conducted on the UCSD ped2, CUHK Avenue, and SHTech Campus datasets, and the results validate the effectiveness of the proposed method.

Document information selection is an essential part of document-grounded dialogue tasks, and more accurate information selection results can provide more appropriate dialogue responses. Existing works have achieved excellent results by employing multi-granularity of dialogue history information, indicating the effectiveness of multi-level historical information. However, these works often focus on exploring the hierarchical information of dialogue history, while neglecting the multi-granularity utilization in response, important information that holds an impact on the decoding process. Therefore, this paper proposes a model for document information selection based on multi-granularity responses. By integrating the document selection results at the response word level and semantic unit level, the model enhances its capability in knowledge selection and produces better responses. For the division at the semantic unit level of the response, we propose two semantic unit division methods, static and dynamic. Experiments on two public datasets show that our models combining static or dynamic semantic unit levels significantly outperform baseline models.

Clustering ensembles can obtain more superior final results by combining multiple different clustering results. The qualities of the points, clusters, and partitions play crucial roles in the consistency of the clustering process. However, existing methods mostly focus on one or two aspects of them, without a comprehensive consideration of the three aspects. This paper proposes a three-level weighted clustering ensemble algorithm namely unified point-cluser-partition algorithm (PCPA). The first step of the PCPA is to generate the adjacency matrix by base clusterings. Then, the central step is to obtain the weighted adjacency matrix by successively weighting three layers, i.e., points, clusters, and partitions. Finally, the consensus clustering is obtained by the average link method. Three performance indexes, namely F, NMI, and ARI, are used to evaluate the accuracy of the proposed method. The experimental results show that: Firstly, as expected, the proposed three-layer weighted clustering ensemble can improve the accuracy of each evaluation index by an average value of 22.07% compared with the direct clustering ensemble without weighting; Secondly, compared with seven other methods, PCPA can achieve better clustering results and the proportion that PCPA ranks first is 28/33.

Hierarchical reinforcement learning (HRL) has achieved remarkable success and significant progress in complex and long-term decision-making problems. However, HRL training typically entails substantial computational costs and an enormous number of samples. One effective approach to tackle this challenge is hierarchical reinforcement learning from demonstrations (HRLfD), which leverages demonstrations to expedite the training process of HRL. The effectiveness of HRLfD is contingent upon the quality of the demonstrations; hence, suboptimal demonstrations may impede efficient learning. To address this issue, this paper proposes a reachability-based reward shaping (RbRS) method to alleviate the negative interference of suboptimal demonstrations for the HRL agent. The novel HRLfD algorithm based on RbRS is named HRLfD-RbRS, which incorporates the RbRS method to enhance the learning efficiency of HRLfD. Moreover, with the help of this method, the learning agent can explore better policies under the guidance of the suboptimal demonstration. We evaluate the proposed HRLfD-RbRS algorithm on various complex robotic tasks, and the experimental results demonstrate that our method outperforms current state-of-the-art HRLfD algorithms.