Brain-tuning enhances brain alignment and downstream performance by fine-tuning speech language models with neural recordings.However, previous work relies primarily on fMRI, whose temporal resolution integrates neural activity over seconds, blending distinct processing stages into a single supervision signal and precluding temporally targeted training.We introduce ECoG-tuning, which leverages electrocorticography's millisecond precision to train speech language models.We design temporally targeted windows-a speech window capturing acoustic-phonetic encoding and a language window capturing higher-order linguistic processing-grounded in neuroscientific findings about temporal encoding hierarchies.Evaluating three models on the Podcast ECoG dataset, we find that ECoG-tuning significantly improves brain alignment over pretrained and distillation baselines.Notably, full spatiotemporal dynamics yield 7-17% higher alignment than time-averaged supervision across models, and language-window tuning produces larger gains in higher-order language regions, indicating that temporal precision provides additional training value.Moreover, ECoG-tuned models consistently improve or maintain downstream performance.Overall, our work provides initial evidence that electrophysiology is a viable brain-tuning modality, demonstrating how neuroscientific insights into processing hierarchies can inform principled model training strategies.

Fault detection in users’ reactive power compensators (URPCs) remains a critical challenge, particularly for general commercial and industrial consumers lacking technical expertise. Undetected URPC malfunctions not only increase electricity costs for users but also aggravate utility power losses. To address this issue, we propose a novel remote fault detection framework that exploits the joint distribution of active load levels and power factors derived from metering data. A vision transformer with a large margin-aware focal model is then employed to effectively classify the operational states of URPCs, using joint frequency distribution matrices as characteristic representations. Unlike conventional approaches, the proposed method relies exclusively on metering data, thereby simplifying deployment and enhancing accessibility for nonspecialist users. This enables timely operation and maintenance of URPCs, reducing electricity costs and improving overall power system efficiency. The effectiveness of the proposed approach is validated through extensive simulations.

Electricity theft users with zero electricity usage (UZEU) should be specifically concerned in electricity theft detection (ETD) research. The challenges are: they provide no effective information on electricity usage behaviors, and they are easily confused with vacant house users. This has caused the majority of the existing detection methods relying on single electricity usage to fail to identify UZEU accurately. Hence, this article first analyzes the underlying correlation between water and electricity (W&E) usage collected by the smart meter. This analysis then lends the theoretical basis to propose a new ETD method by comprehensively using the multisource information. More precisely, the proposed method utilizes the mutual information coefficient (MIC) to construct a correlation model between W&E usage and in turn the wavelet clustering algorithm to cluster the MIC of the power distribution users. Thereafter, the resulting weak correlations indicate the suspected users as the electricity theft UZEU in case of zero electricity usage. Finally, the proposed method is validated by numerical experiments in the real world and illustrated to be more accurate than existing methods in detecting UZEU.