Yi Gu

Yi Gu, Huaiguang Jiang, Jun Jason Zhang et al.

In this paper, a short-term load forecasting approach based network reconfiguration is proposed in a parallel manner. Specifically, a support vector regression (SVR) based short-term load forecasting approach is designed to provide an accurate load prediction and benefit the network reconfiguration. Because of the nonconvexity of the three-phase balanced optimal power flow, a second-order cone program (SOCP) based approach is used to relax the optimal power flow problem. Then, the alternating direction method of multipliers (ADMM) is used to compute the optimal power flow in distributed manner. Considering the limited number of the switches and the increasing computation capability, the proposed network reconfiguration is solved in a parallel way. The numerical results demonstrate the feasible and effectiveness of the proposed approach.

Yi Gu, Huaiguang Jiang, Jun Jason Zhang et al.

This paper aims to propose a two-step approach for day-ahead hourly scheduling in a distribution system operation, which contains two operation costs, the operation cost at substation level and feeder level. In the first step, the objective is to minimize the electric power purchase from the day-ahead market with the stochastic optimization. The historical data of day-ahead hourly electric power consumption is used to provide the forecast results with the forecasting error, which is presented by a chance constraint and formulated into a deterministic form by Gaussian mixture model (GMM). In the second step, the objective is to minimize the system loss. Considering the nonconvexity of the three-phase balanced AC optimal power flow problem in distribution systems, the second-order cone program (SOCP) is used to relax the problem. Then, a distributed optimization approach is built based on the alternating direction method of multiplier (ADMM). The results shows that the validity and effectiveness method.

Yi Gu

Due to the large volume of heterogeneous data provided by both the customer and the grid side, a big data visualization platform is built to discover the hidden useful knowledge for smart grid (SG) operation, control and situation awareness. An open source cluster computing framework based on Apache Spark is considering to discover the hidden knowledge of the bag-data. And the data is transmitted with a Open System Interconnection (OSI) model to the data visualization platform in a high-speed communication architecture. Google Earth and global geographic information system (GIS) are used to design the visualization platform and realize the result with the test bench. The alternating direction method of multipliers (ADMM) is used to compute the optimal power flow in distributed manner. The proposed network reconfiguration is solved in parallel manner with the limited switches and the strong computing capability. Further more, a multi-timescale operation approach is modeled with a three-phase distributed system, which consists the hourly scheduling at substation level and the minutes power flow operation at feeder level. The system cost with renewable generation is minimized at the substation level. The given error distribution model of renewable generation is simulated with a chance constraint, and the derived deterministic form is modeled with Gaussian mixture model (GMM) with genetic algorithm-based expectationmaximization (GAEM). The system cost is further reduced with the OPF in real-time (RT) scheduling. The semidefinite programming (SDP) is used to relax the nonconvexity of the three-phase unbalanced distribution system into a convex problem, which make sure to achieve the global optimal result. In the parallel manner, the ADMM is realizing getting the results in a short time.

Yi Gu, Yoshito Otake, Keisuke Uemura et al.

While most vision tasks are essentially visual in nature (for recognition), some important tasks, especially in the medical field, also require quantitative analysis (for quantification) using quantitative images. Unlike in visual analysis, pixel values in quantitative images correspond to physical metrics measured by specific devices (e.g., a depth image). However, recent work has shown that it is sometimes possible to synthesize accurate quantitative values from visual ones (e.g., depth from visual cues or defocus). This research aims to improve quantitative image synthesis (QIS) by exploring pretraining and image resolution scaling. We propose a benchmark for evaluating pretraining performance using the task of QIS-based bone mineral density (BMD) estimation from plain X-ray images, where the synthesized quantitative image is used to derive BMD. Our results show that appropriate pretraining can improve QIS performance, significantly raising the correlation of BMD estimation from 0.820 to 0.898, while others do not help or even hinder it. Scaling-up the resolution can further boost the correlation up to 0.923, a significant enhancement over conventional methods. Future work will include exploring more pretraining strategies and validating them on other image synthesis tasks.

Letian Peng, Yi Gu, Chengyu Dong et al.

For extremely weak-supervised text classification, pioneer research generates pseudo labels by mining texts similar to the class names from the raw corpus, which may end up with very limited or even no samples for the minority classes. Recent works have started to generate the relevant texts by prompting LLMs using the class names or definitions; however, there is a high risk that LLMs cannot generate in-distribution (i.e., similar to the corpus where the text classifier will be applied) data, leading to ungeneralizable classifiers. In this paper, we combine the advantages of these two approaches and propose to bridge the gap via a novel framework, \emph{text grafting}, which aims to obtain clean and near-distribution weak supervision for minority classes. Specifically, we first use LLM-based logits to mine masked templates from the raw corpus, which have a high potential for data synthesis into the target minority class. Then, the templates are filled by state-of-the-art LLMs to synthesize near-distribution texts falling into minority classes. Text grafting shows significant improvement over direct mining or synthesis on minority classes. We also use analysis and case studies to comprehend the property of text grafting.

Shibo Hao, Yi Gu, Haodi Ma et al.

Large language models (LLMs) have shown remarkable reasoning capabilities, especially when prompted to generate intermediate reasoning steps (e.g., Chain-of-Thought, CoT). However, LLMs can still struggle with problems that are easy for humans, such as generating action plans for executing tasks in a given environment, or performing complex math, logical, and commonsense reasoning. The deficiency stems from the key fact that LLMs lack an internal $\textit{world model}$ to predict the world $\textit{state}$ (e.g., environment status, intermediate variable values) and simulate long-term outcomes of actions. This prevents LLMs from performing deliberate planning akin to human brains, which involves exploring alternative reasoning paths, anticipating future states and rewards, and iteratively refining existing reasoning steps. To overcome the limitations, we propose a new LLM reasoning framework, $\underline{R}$easoning vi$\underline{a}$ $\underline{P}$lanning $\textbf{(RAP)}$. RAP repurposes the LLM as both a world model and a reasoning agent, and incorporates a principled planning algorithm (based on Monto Carlo Tree Search) for strategic exploration in the vast reasoning space. During reasoning, the LLM (as agent) incrementally builds a reasoning tree under the guidance of the LLM (as world model) and task-specific rewards, and obtains a high-reward reasoning path efficiently with a proper balance between exploration $\textit{vs.}$ exploitation. We apply RAP to a variety of challenging reasoning problems including plan generation, math reasoning, and logical inference. Empirical results on these tasks demonstrate the superiority of RAP over various strong baselines, including CoT and least-to-most prompting with self-consistency. RAP on LLAMA-33B surpasses CoT on GPT-4 with 33% relative improvement in a plan generation setting.

Jiannan Xiang, Tianhua Tao, Yi Gu et al.

While large language models (LMs) have shown remarkable capabilities across numerous tasks, they often struggle with simple reasoning and planning in physical environments, such as understanding object permanence or planning household activities. The limitation arises from the fact that LMs are trained only on written text and miss essential embodied knowledge and skills. In this paper, we propose a new paradigm of enhancing LMs by finetuning them with world models, to gain diverse embodied knowledge while retaining their general language capabilities. Our approach deploys an embodied agent in a world model, particularly a simulator of the physical world (VirtualHome), and acquires a diverse set of embodied experiences through both goal-oriented planning and random exploration. These experiences are then used to finetune LMs to teach diverse abilities of reasoning and acting in the physical world, e.g., planning and completing goals, object permanence and tracking, etc. Moreover, it is desirable to preserve the generality of LMs during finetuning, which facilitates generalizing the embodied knowledge across tasks rather than being tied to specific simulations. We thus further introduce the classical (EWC) for selective weight updates, combined with low-rank adapters (LoRA) for training efficiency. Extensive experiments show our approach substantially improves base LMs on 18 downstream tasks by 64.28% on average. In particular, the small LMs (1.3B, 6B, and 13B) enhanced by our approach match or even outperform much larger LMs (e.g., ChatGPT).