Zheng Li

Chain-of-Thought (CoT) reasoning, with long or short variants, has become a central paradigm for enhancing the reasoning ability of large language models. Long CoT aims at achieving high accuracy in complex multi-step reasoning at the cost of high inference overhead, while short CoT focus on offering a more efficient but less capable alternative. To balance reasoning accuracy and reasoning efficiency, collaborative approaches combining long and short CoT have been widely explored. However, existing collaborative approaches either operate token-level pruning, risking disruption to reasoning coherence, or rely on instance-level routing, which over-look the excessive self-verification within long CoT reasoning. To address these limitations, a semantic step-level reasoning optimization approach for collaborative CoTs, named SemCoT, is investigated in this paper. Technically, verification redundancy and trigger mechanisms in long CoT reasoning are first analyzed. Specifically, a dynamic trigger leveraging lexical cues and logit uncertainty is designed to detect the onset of self-verification in long CoT reasoning. A finetuned short CoT model is subsequently employed to filter trivial verification, with redundant cases truncated through context steering. Confident phrases are finally embedded to ensure contextual coherence. Experimental results demonstrate that SemCoT reduces average reasoning length by 48% across several benchmarks while maintaining similar accuracy, highlighting its practical effectiveness.

To improve solar irradiance simulation accuracy for precise photovoltaic power forecasting, we developed a hybrid framework combining WRF-Solar numerical simulation and random forest (RF) machine learning for a PV plant in Guangdong, China. Weather conditions were objectively classified into clear, intermittent cloudy, and overcast using the Daily Variability Index (DVI) and Daily Clear-sky Index (DCI). We calibrated the WRF-Solar model’s microphysics and radiative transfer schemes via sensitivity tests to optimize overcast-sky performance, then applied RF correction to the simulated irradiance. Results show that RF correction significantly reduces simulation errors for intermittent and overcast conditions, while the original WRF-Solar outperforms the corrected results under clear skies due to RF overfitting.

This study simulated the wind energy density distribution in the Jiaodong Peninsula region using the Weather Research and Forecasting (WRF) Model. The impacts of different boundary-layer and near-surface parameterization schemes on the simulated wind speed and direction were investigated. The results indicate that the Yonsei University (YSU) scheme and the Quasi-Normal Scale Elimination (QNSE) scheme performed optimally for wind speed and wind direction. We also conducted a sensitivity test of the simulation results for atmospheric pressure, air temperature, and relative humidity. The statistical analysis showed that the YSU scheme performed optimally, while the MRF and BL schemes performed poorly. Following this, the wind energy distribution in the coastal hilly areas of the Jiaodong Peninsula was simulated using the YSU boundary-layer parameterization scheme. The modeled wind energy density in the mountainous and hilly areas of the Jiaodong Peninsula were higher than that in other regions. The wind energy density exhibits a seasonal variation, with the highest values in spring and early summer and the lowest in summer. In spring, the wind energy density over the Bohai Sea is higher than over the Yellow Sea, while the opposite trend is modeled in summer.

This study simulated the spatial distribution of wind speeds and wind energy density by using the WRF model, and we used the WRF-simulated results to evaluate the sea surface wind speeds retrieved from the HY-2C and CFOSAT satellite-borne microwave scatterometers over the Yellow Sea region. The main conclusions were as follows: (1) The combination of the MRF boundary layer parameterization scheme, the MM5 near-surface parameterization scheme, and the Global Data Assimilation System (GDAS) initial field demonstrated the best performance in simulating the 10 m wind speed in the Yellow Sea region, with a root-mean-square error (RMSE) of 1.57, bias of 1.24 m/s, and mean absolute percentage error (MAPE) of 17%. (2) The MAPE of the HY-2C inversion data was 9%, while the CFOSAT inversion data had an MAPE of 6%. The sea surface wind speeds derived from the HY-2C and CFOSAT satellite scatterometer inversions demonstrated high accuracy and applicability in this region. (3) The wind speed was found to increase with altitude over the Yellow Sea, with higher wind speeds observed in the southern region compared to the northern region. The wind power density increased with altitude, and the wind power density in the southern area of the Yellow Sea was higher than in the northern region. (4) The CFOSAT satellite inversion products were in good agreement with the WRF simulation results under low wind speed conditions. In contrast, the HY-2C satellite inversion products showed better agreement under moderate wind speed conditions. Under high wind speed conditions, both satellite inversion products exhibited minor deviations, but the HY-2C product had an overall overestimation, while the CFOSAT product remained within the range of −1 to 1 m/s. (6) The wind power density increased with the satellite-inverted 10 m wind speed. When the 10 m wind speed was less than 9 m/s, the wind power density exhibited a roughly cubic trend of increase. However, when the 10 m wind speed exceeded 9 m/s, the wind power density no longer increased with the rise in 10 m wind speed. These findings provide valuable insights into wind energy resources in the Yellow Sea region and demonstrate the effectiveness of satellite scatterometer inversions for wind speed estimation. The results have implications for renewable energy planning and management in the area.

The distribution of edge resources in the edge computing environment has an important impact on the Quality of Service (QoS) of edge services. Unreasonable server placement will inevitably lead to problems such as server overload or underload, deteriorating workload balancing and service wait time. Therefore, the key issue to be addressed in server placement is how to enhance the QoS of edge services through efficient edge server placement strategies under multiple requirements such as average task wait time and data privacy. Edge computing (EC) assisted with blockchain technology was argued to be the most potential solution. In this paper, we propose a blockchain-assisted secure ES placement algorithm named ETS_GA. ETS_GA is based on the elite-preserving genetic algorithm (EGA), which is proven to converge. The premature problem of traditional genetic algorithm (GA) is effectively solved by using tabu search (TS) and niche sharing (NS). In addition, we construct an adaptive state supervising machine (ASM) to realize real-time algorithm supervision and adaptively iterate the optimization strategy. Blockchain-based privacy protection methods are also deployed in the placed servers to provide real-time privacy protection. Finally, our proposed method is experimentally compared with four baselines using the real Shanghai Telecom base station data set, whose results demonstrate the superiority of ETS_GA in terms of convergence and global search capability.

In the social production system, image data are rapidly generated from almost all fields such as factories, hospitals, and transportation, promoting higher requirements for image anomaly detection technologies, including low consumption, higher adaptability, and accuracy. However, existing anomaly detection methods are fragile to heterogeneous image data generated by complex social production systems and tend to require strong computing power and resource support. To address the above problems, a knowledge-driven anomaly detection framework is proposed, in which a local feature enhancement method is designed to strengthen the knowledge representation of the initial features extracted from images. The attention mechanism in deep learning is introduced to adjust the feature attention dynamically according to prior knowledge, which solves the problem of feature loss in the cascade training. To verify the effectiveness of the proposed framework, extensive experiments on social production datasets are conducted. The results demonstrate that our framework outperforms the selected methods on image datasets with different complexity and sample distributions.

Recommender systems are technology-driven marketing solutions for businesses that analyze user behavior data. However, collaborative data sharing between enterprises is often prohibited by privacy protection regulations, leading to insufficient data for graph neural networks (GNNs) training. Fortunately, federated learning (FL), a collaborative training framework without exposing source data, can be applied congruently. Nevertheless, most of FL-based GNN model training methods adopt federated averaging, which performs poorly on highly heterogeneous graph data. To solve this problem, a FL-based GNN Model Training framework for cross-enterprise recommendation, named FL-GMT, is proposed. Specifically, a GNN-based recommendation model is deployed as the local training model. Then, considering the performance inequity caused by uneven sample quality, a loss-based federated aggregation algorithm is designed, effectively improving the performance of disadvantaged participants. To improve the system stability at the end of the aggregation, a dynamic update method of loss attention is designed. Extensive experiments on benchmark datasets demonstrate that FL-GMT outperforms baselines in terms of system fairness, stability, and accuracy.

With the increasing data scale in the Industrial Internet of Things, edge computing coordinated with machine learning is regarded as an effective way to raise the novel latency-sensitive services. To ensure the data privacy for frequent service access, federated learning (FL), as a privacy-preserving distributed framework, is integrated into edge computing, enabling user data invisible to the training process. However, sophisticated network attacks threaten deep learning (DL) models by data poison and malicious reasoning, making the DL-based system untrustworthy. To this end, a synergic data filtering method, named Safe, is proposed to deal with the poisoning attacks. Specifically, considering that the distributed support vector machine is at risk of being attacked due to its distribution and openness to communication, edge-cloud empowered FL framework is designed. Then, the alternating direction method of multipliers is deployed to detect attacked devices whose training processes will be interrupted. Moreover, due to the untrustworthiness of label data, the poisoned data in the attacked devices are figured out and filtered by clustering the trusted data with K -means clustering algorithm. Eventually, extensive experiment results proved that the Safe outperforms correlation methods in detection accuracy and trustworthiness.