Large language models are extensively utilized in creative writing applications. Creative writing requires a balance between subjective writing quality (e.g., literariness and emotional expression) and objective constraint following (e.g., format requirements and word limits). Existing reinforcement learning methods struggle to balance these two aspects: single reward strategies fail to improve both abilities simultaneously, while fixed-weight mixed-reward methods lack the ability to adapt to different writing scenarios. To address this problem, we propose Reinforcement Learning with Mixed Rewards (RLMR), utilizing a dynamically mixed reward system from a writing reward model evaluating subjective writing quality and a constraint verification model assessing objective constraint following. The constraint following reward weight is adjusted dynamically according to the writing quality within sampled groups, ensuring that samples violating constraints get negative advantage in GRPO and thus penalized during training, which is the key innovation of this proposed method. We conduct automated and manual evaluations across diverse model families from 8B to 72B parameters. Additionally, we construct a real-world writing benchmark named WriteEval for comprehensive evaluation. Results illustrate that our method achieves consistent improvements in both instruction following (IFEval from 83.36% to 86.65%) and writing quality (72.75% win rate in manual expert pairwise evaluations on WriteEval). To the best of our knowledge, RLMR is the first work to combine subjective preferences with objective verification in online RL training, providing an effective solution for multi-dimensional creative writing optimization.

The spread of rumors on social media, particularly during significant events like the US elections and the COVID-19 pandemic, poses a serious threat to social stability and public health. Current rumor detection methods primarily rely on propagation graphs to improve the model performance. However, the effectiveness of these methods is often compromised by noisy and irrelevant structures in the propagation process. To tackle this issue, techniques such as weight adjustment and data augmentation have been proposed. However, they depend heavily on rich original propagation structures, limiting their effectiveness in handling rumors that lack sufficient propagation information, especially in the early stages of dissemination. In this work, we introduce the Key Propagation Graph Generator (KPG), a novel reinforcement learning-based framework, that generates contextually coherent and informative propagation patterns for events with insufficient topology information and identifies significant substructures in events with redundant and noisy propagation structures. KPG comprises two key components: the Candidate Response Generator (CRG) and the Ending Node Selector (ENS). CRG learns latent variable distributions from refined propagation patterns to eliminate noise and generate new candidates for ENS, while ENS identifies the most influential substructures in propagation graphs and provides training data for CRG. Furthermore, we develop an end-to-end framework that utilizes rewards derived from a pre-trained graph neural network to guide the training process. The resulting key propagation graphs are then employed in downstream rumor detection tasks. Extensive experiments conducted on four datasets demonstrate that KPG outperforms current state-of-the-art methods.

The less developed sensor technologies of traditional civil elevators, particularly in older districts, have long impeded the construction of smart cities. By using surveillance cameras, we can intelligently identify the abnormal events in elevator, but there is currently a lack of relevant datasets. For this purpose, we curate a visual elevator anomalous events dataset that serves as a benchmark for evaluating the performance of state-of-the-art detection models like YOLOv8. Then we improved YOLOv8 via adding coordinate attention and optimizing loss functions and activation functions, resulting in a notable enhancement in its performance for elevator abnormal detection. Furthermore, based on the improved model we implement a system called ELEvent in a real residential community and proved effective in practice.

Given a graph$\mathcal {G}$with a set of attributes, theattributed network embedding (ANE)aims to learn low-dimensional representations of nodes that preserve both graph topology and node attribute proximity. ANE is shown to be more effective than plain network embedding methods (using only graph topology) on many graph mining tasks. However, existing ANE solutions still provide inferior performance on tasks like node classification and link prediction, as will be shown in our experiments. The key issue is that when combining graph topology and attribute information, most existing solutions take attributes with equal importance, while in real scenarios, different attribute exerts distinct influence over the network due to the heterogeneous nature among attributes. Motivated by this, we presentLATAM, a learning-based framework for ANE via trainable proximity matrix factorization. To capture the node-attribute relationships, we first construct the attribute-augmented graph by adding attribute nodes (resp. edges) to the original graph. Then, we define the attribute-augmented random walk and proximity on the attribute-augmented graph, where the weights of different attributes can be learned automatically by our designed loss functions so that more indicative attributes tend to have higher weights, imposing a higher impact on the node connectivity. To achieve this, we incorporate a differentiable SVD to back-propagate gradients of attribute weights in an end-to-end process. To scale our LATAM to large graphs, we further propose sampling techniques to learn attribute weights and an efficient attribute-augmented push algorithm to compute the proximity matrix. Extensive experiments on 8 public attributed networks against 11 existing methods show the effectiveness of our LATAM.

Control data separation architecture (CDSA) is a more efficient architecture to overcome the overhead issue than the conventional cellular networks, especially for the huge bursty traffic like Internet of Things, and over-the-top (OTT) content service. In this paper, we study the optimization issue of network energy efficiency of the CDSA-based heterogeneous cloud radio access networks (H-CRAN) networks, which has heterogeneous fronthaul between control base station (CBS) and data base stations (DBSs). We first present a modified power consumption model for the CDSA-based H-CRAN, and then formulate the optimization problem with constraint of overall capacity of wireless fronthaul. Then we work out the resource assignment and power allocation by the convex relaxation approach using fractional programming, norm approximation, and Lagrangian dual decomposition method, with a derived the close-form optimal solution. Finally, we verify the proposed method by system- level simulation. The comprehensive simulation results show that our proposed algorithm has 10% EE gain compared to the static algorithm, and the CDSA-based H-CRAN networks can achieve up to 14% EE gain compared to the conventional network even under strict fronthaul capacity limit.

The Data Distribution Management (DDM) service in the runtime infrastructure of high level architecture is employed to reduce the amount of irrelevant data communicated between federates and cut network communication cost. Current DDM schemes include mainly region-based and grid-based filtering mechanisms. We propose an agent-based DDM filtering mechanism to use agents to perform accurate data filtering in DDM. The agent-based DDM is supposed to filter out all the irrelevant information and route only the information that subscribers want. In this paper, an agent-based DDM mechanism is developed and incorporated into the runtime infrastructure. The efficiencies of the agent-based filtering mechanism and the other two traditional mechanisms are compared in an AWACS sensing aircraft experiment scenario. The results show that agent-based filtering mechanism communicates only the exactly relevant data and minimizes the network communication cost.