Quantifying image complexity at the entity level is straightforward, but the assessment of semantic complexity has been largely overlooked. In fact, there are differences in semantic complexity across images. Images with richer semantics can tell vivid and engaging stories and offer a wide range of application scenarios. For example, the Cookie Theft picture is such a kind of image and is widely used to assess human language and cognitive abilities due to its higher semantic complexity. Additionally, semantically rich images can benefit the development of vision models, as images with limited semantics are becoming less challenging for them. However, such images are scarce, highlighting the need for a greater number of them. For instance, there is a need for more images like Cookie Theft to cater to people from different cultural backgrounds and eras. Assessing semantic complexity requires human experts and empirical evidence. Automatic evaluation of how semantically rich an image will be the first step of mining or generating more images with rich semantics, and benefit human cognitive assessment, Artificial Intelligence, and various other applications. In response, we propose the Image Semantic Assessment (ISA) task to address this problem. We introduce the first ISA dataset and a novel method that leverages language to solve this vision problem. Experiments on our dataset demonstrate the effectiveness of our approach.

Multivariate time series (MTS) forecasting is crucial in the finance, energy and transportation sectors. Although current models deliver high prediction accuracy, their complexity and extensive computational demands hinder practical deployment. To address these limitations, this paper introduces a lightweight multiscale spatio-temporal convolutional network (LW-MSTCNN) that leverages deep separable convolution (DSCNN) to build an efficient multiscale pyramid model to decrease model complexity and computational costs. Additionally, a multiscale attention (MA) mechanism is incorporated to selectively emphasize important features across different spatial and temporal scales, improving feature extraction and overall performance. This combined approach enhances the model’s ability to process high-dimensional data efficiently while capturing critical multiscale patterns, addressing key challenges in spatio-temporal data processing. Experimental results on four datasets demonstrate that the proposed model significantly reduces the usage of computational resources, with the parameter count reduced by 4 to 5 times, while maintaining comparable prediction accuracy. Additionally, extensive generalization experiments show that the model exhibits strong robustness. This highlights the effectiveness and practicality of the model in multivariate time series (MTS) forecasting.

The security of virtual power plant (VPP) communication networks is paramount, particularly owing to growing increase in the reliance on distributed energy resources (DERs) using cloud-edge architectures. VPPs are increasingly vulnerable to cyber intrusion owing to the large number of access devices involved. Existing intrusion detection methods often face challenges in addressing the complexity of VPP networks, and lack adaptability to diverse attack patterns. To address the dynamic nature of VPPs, this study proposes a novel intrusion detection approach called dynamic detection combining transformer with feature filtering (DTF-VPP), which integrates principal component analysis (PCA) with a transformer-based model to improve the detection efficiency and accuracy. The key contributions of this study include a feature selection process that uses PCA to reduce the model complexity, and a dynamic loss-weighting mechanism that adapts to high-frequency attacks. The experimental results on the NSL-KDD dataset demonstrate that DTF-VPP outperforms conventional models in terms of the accuracy and F1-score. Therefore, this approach offers a scalable and adaptive solution for enhancing the security of VPPs against cyber threats.

Curriculum learning has shown promising improvements in multiple domains by training machine learning models from easy samples to hard ones.Previous works which either design rules or train models for scoring the difficulty highly rely on task-specific expertise, and cannot generalize.Inspired by the "easy-to-hard" intuition, we propose to do in-sample curriculum learning for natural language generation tasks.Our learning strategy starts training the model to generate the last few words, i.e., do sequence completion, and gradually extends to generate the whole output sequence.Comprehensive experiments show that it generalizes well to different tasks and achieves significant improvements over strong baselines.

Existing automatic evaluation systems of chatbots mostly rely on static chat scripts as ground truth, which is hard to obtain, and requires access to the models of the bots as a form of "white-box testing".Interactive evaluation mitigates this problem but requires human involvement.In our work, we propose an interactive chatbot evaluation framework in which chatbots compete with each other like in a sports tournament, using flexible scoring metrics.This framework can efficiently rank chatbots independently from their model architectures and the domains for which they are trained.

Audio tagging aims at predicting sound events occurred in a recording. Traditional models require enormous laborious annotations, otherwise performance degeneration will be the norm. Therefore, we investigate robust audio tagging models in low-resource scenarios with the enhancement of knowledge graphs. Besides existing ontological knowledge, we further propose a semi-automatic approach that can construct temporal knowledge graphs on diverse domain-specific label sets. Moreover, we leverage a variant of relation-aware graph neural network, D-GCN, to combine the strength of the two knowledge types. Experiments on AudioSet and SONYC urban sound tagging datasets suggest the effectiveness of the introduced temporal knowledge, and the advantage of the combined KGs with D-GCN over single knowledge source.

Multi-turn response selection is a task designed for developing dialogue agents.The performance on this task has a remarkable improvement with pre-trained language models.However, these models simply concatenate the turns in dialogue history as the input and largely ignore the dependencies between the turns.In this paper, we propose a dialogue extraction algorithm to transform a dialogue history into threads based on their dependency relations.Each thread can be regarded as a self-contained sub-dialogue.We also propose Thread-Encoder model to encode threads and candidates into compact representations by pre-trained Transformers and finally get the matching score through an attention layer.The experiments show that dependency relations are helpful for dialogue context understanding, and our model outperforms the state-of-the-art baselines on both DSTC7 and DSTC8*, with competitive results on UbuntuV2.

Modern high-end scientific and technological robots have been utilized in various fields. With regard to distribution, reliability, flexibility, and economy, swarm robots can accomplish more arduous tasks in parallel and demonstrate a superior performance compared to single robot. However, information interaction, which causes extra time overhead, is essential to realize synchronization or coordination in multi robot system. This paper focuses on improving the edge coverage ability of base stations especially when swarm robots cover a wide range. It proposes a novel collaborative communication protocol named Co2-Robot for these robots in order to carry out distributed beamforming, thus enhancing the coverage range. In particular, efficient strategies are designed to cut down the time overhead among swarm robots. Subsequently, we develop a novel cluster head selection scheme to be responsible for the parallel operations. Then, distributed beamforming is performed to improve the uplink coverage when no signal from anyone of swarm robots can reach the base station. In addition, several practical user-cases are considered and analyzed in distributed beamforming, in order to enhance the robustness of this parallel system. Finally, extensive simulations show that Co2-Robot will just result in a small amount of time overhead, which is acceptable in most scenarios. Furthermore, it can substantially extend the coverage range up to nearly 1000m while single robot can just cover a coverage range of more than 100m.