arXiv subjects

Jing Shang

Publications and source records attributed to Jing Shang.

9 recordsLinked to original sources

arXiv2026

HERO: Human-profile Enhanced Retrieval Optimization Framework for Long-term Agent Memory

Long-term memory is crucial for personalized responses and long-horizon agent interactions. Existing methods often rely on LLMs to compress or rewrite dialogue histories and use the transformed memories as retrieval evidence. Despite the progress in organizing fragmented contexts, two major drawbacks persist: (1) information loss from compression, which discards fine-grained but later useful details, and (2) semantic drift from rewriting, which erodes the original tone and situated context. In this work, we propose a novel Human-profile Enhanced Retrieval Optimization framework for long-term agent memory (HERO). Specifically, HERO converts the dialogue history into a traceable heterogeneous memory graph that preserves raw dialogue text as evidence for reasoning, thereby mitigating information loss. For retrieval, HERO extracts initial anchors from the current query and incorporates human profiles via an iterative graph traversal; these anchors and profiles provide guidance signals that adaptively activate the most informative regions of the graph. Experiments on two benchmark datasets show that HERO outperforms strong baselines on both factual and personalized reasoning, while providing more faithful access to raw dialogue evidence.

cs.AI↗

arXiv2026

Unlocking Deep Demand Flexibility via Dynamic Signals

The rapid proliferation of distributed energy resources (DERs) and the electrification of residential loads offer significant potential for grid flexibility but pose stability challenges under static pricing regimes. Specifically, high levels of automation under static Time-of-Use (TOU) tariffs often induce ``device synchronization,'' where simultaneous responses from home energy management systems (HEMS) create artificial demand peaks that threaten grid stability. This paper proposes a privacy-preserving, one-way dynamic signaling framework to unlock deep demand flexibility from HEMS. We utilize a feedback-based learning algorithm that updates day-ahead price profiles based on aggregate substation demand and environmental contexts, effectively closing the loop between utility objectives and aggregated edge behaviors. The framework is rigorously validated using high-fidelity simulations on an 84-bus distribution network populated with hundreds of HEMS controlling diverse devices, including HVAC, PV, batteries, and flexible loads. Results demonstrate that the proposed mechanism achieves substantial reductions in both peak demand and total load variation. Extensive analyses across diverse climates and scalable deployments confirm the framework's robustness, indicating that dynamic pricing acts as a force multiplier for DERs, with peak shaving potential increasing significantly under high renewable penetration scenarios.

math.OC↗

arXiv2025

Lassoed Forests: Random Forests with Adaptive Lasso Post-selection

Random forests are a statistical learning technique that use bootstrap aggregation to average high-variance and low-bias trees. Improvements to random forests, such as applying Lasso regression to the tree predictions, have been proposed in order to reduce model bias. However, these changes can sometimes degrade performance (e.g., an increase in mean squared error). In this paper, we show in theory that the relative performance of these two methods, standard and Lasso-weighted random forests, depends on the signal-to-noise ratio. We further propose a unified framework to combine random forests and Lasso selection by applying adaptive weighting and show mathematically that it can strictly outperform the other two methods. We compare the three methods through simulation, including bias-variance decomposition, error estimates evaluation, and variable importance analysis. We also show the versatility of our method by applications to a variety of real-world datasets.

stat.ML↗

arXiv2025

Pre-validation Revisited

Pre-validation is a way to build prediction model with two datasets of significantly different feature dimensions. Previous work showed that the asymptotic distribution of the resulting test statistic for the pre-validated predictor deviates from a standard Normal, hence leads to issues in hypothesis testing. In this paper, we revisit the pre-validation procedure and extend the problem formulation without any independence assumption on the two feature sets. We propose not only an analytical distribution of the test statistic for the pre-validated predictor under certain models, but also a generic bootstrap procedure to conduct inference. We show properties and benefits of pre-validation in prediction, inference and error estimation by simulations and applications, including analysis of a breast cancer study and a synthetic GWAS example.

stat.ME↗

arXiv2024

Improve ROI with Causal Learning and Conformal Prediction

In the commercial sphere, such as operations and maintenance, advertising, and marketing recommendations, intelligent decision-making utilizing data mining and neural network technologies is crucial, especially in resource allocation to optimize ROI. This study delves into the Cost-aware Binary Treatment Assignment Problem (C-BTAP) across different industries, with a focus on the state-of-the-art Direct ROI Prediction (DRP) method. However, the DRP model confronts issues like covariate shift and insufficient training data, hindering its real-world effectiveness. Addressing these challenges is essential for ensuring dependable and robust predictions in varied operational contexts. This paper presents a robust Direct ROI Prediction (rDRP) method, designed to address challenges in real-world deployment of neural network-based uplift models, particularly under conditions of covariate shift and insufficient training data. The rDRP method, enhancing the standard DRP model, does not alter the model's structure or require retraining. It utilizes conformal prediction and Monte Carlo dropout for interval estimation, adapting to model uncertainty and data distribution shifts. A heuristic calibration method, inspired by a Kaggle competition, combines point and interval estimates. The effectiveness of these approaches is validated through offline tests and online A/B tests in various settings, demonstrating significant improvements in target rewards compared to the state-of-the-art method.

cs.LG↗

arXiv2023

Ferroelectric Domain and Switching Dynamics in Curved In2Se3: First Principle and Deep Learning Molecular Dynamics Simulations

Complex strain status can exist in 2D materials during their synthesis process, resulting in significant impacts on the physical and chemical properties. Despite their prevalence in experiments, their influence on the material properties and the corresponding mechanism are often understudied due to the lack of effective simulation methods. In this work, we investigated the effects of bending, rippling, and bubbling on the ferroelectric domains in In2Se3 monolayer by density functional theory (DFT) and deep learning molecular dynamics (DLMD) simulations. The analysis of the tube model shows that bending deformation imparts asymmetry into the system, and the polarization direction tends to orient towards the tensile side, which has a lower energy state than the opposite polarization direction. The energy barrier for polarization switching can be reduced by compressive strain according DFT results. The dynamics of the polarization switching is investigated by the DLMD simulations. The influence of curvature and temperature on the switching time follows the Arrhenius-style function. For the complex strain status in the rippling and bubbling model, the lifetime of the local transient polarization is analyzed by the autocorrelation function, and the size of the stable polarization domain is identified. Local curvature and temperature can influence the local polarization dynamics following the proposed Arrhenius-style equation. Through cross-scale simulations, this study demonstrates the capability of deep-learning potentials in simulating polarization for ferroelectric materials. It further reveals the potential to manipulate local polarization in ferroelectric materials through strain engineering.

cond-mat.mtrl-sci↗

arXiv2020

Multiferroic Decorated Fe2O3 Monolayer Predicted from First Principles

Two-dimensional (2D) multiferroics exhibit cross-control capacity between magnetic and electric responses in reduced spatial domain, making them well suited for next-generation nanoscale devices; however, progress has been slow in developing materials with required characteristic properties. Here we identify by first-principles calculations robust 2D multiferroic behaviors in decorated Fe2O3 monolayer, showcasing N@Fe2O3 as a prototypical case, where ferroelectricity and ferromagnetism stem from the same origin, namely Fe d-orbit splitting induced by the Jahn-Teller distortion and associated crystal field changes. The resulting ferromagnetic and ferroelectric polarization can be effectively reversed and regulated by applied electric field or strain, offering efficient functionality. These findings establish strong materials phenomena and elucidate underlying physics mechanism in a family of truly 2D multiferroics that are highly promising for advanced device applications.

cond-mat.mtrl-sci↗

arXiv2020

Robust Magnetoelectric Effect in Decorated Graphene/In2Se3 Heterostructure

Magnetoelectric effect is a fundamental physics phenomenon that synergizes electric and magnetic degrees of freedom to generate distinct material responses like electrically tuned magnetism, which serves as a key foundation of the emerging field of spintronics. Here, we show by first-principles studies that ferroelectric (FE) polarization of an In2Se3 monolayer can modulate the magnetism of an adjacent transition-metal (TM) decorated graphene layer via an FE induced electronic transition. The TM nonbonding d-orbital shifts downward and hybridizes with carbon p states near the Fermi level, suppressing the magnetic moment, under one FE polarization, but on reversed FE polarization this TM d-orbital moves upward, restoring the original magnetic moment. This finding of robust magnetoelectric effect in TM decorated graphene/In2Se3 heterostructure offers powerful insights and a promising avenue for experimental exploration of FE controlled magnetism in 2D materials.

cond-mat.mtrl-sci↗

arXiv2019

Reversible Gas Sensing by Ferroelectric Switch and 2D Molecule Multiferroics in In2Se3 Monolayer

Two-dimensional ferroelectrics are important quantum materials which have found novel application in nonvolatile memories, however, the effects of reversible polarization on chemical reactions and interaction with environments are rarely studied despite of its importance. Here, based on the first-principles calculations, we found distinct gas adsorption behaviors on the surfaces of ferroelectric In2Se3 layer and the reversible gas caption and release controlled by ferroelectric switch. We rationalize the novel phenomena to the synergistic effect of the different electrostatic potential and electron transfer induced by band alignments between frontier molecular orbitals of gas and band edge states of substrate. Excitingly, the adsorption of paramagnetic gas molecules such as NO and NO2 can induce surface magnetism, which is also sensitive to ferroelectric polarization direction of In2Se3, indicating the application of In2Se3 as threshold magnetic sensors or switcher. Furthermore, it is suggested two NO molecules prefer to ferromagnetically couple with each other, the Curie temperature is polarization dependent which can reach up to 50K, leading to the long-sought 2D molecule multiferroics. The ferroelectric controllable adsorption behavior and molecule multiferroic feature will find extensive application in gas caption, selective catalytic reduction and spintronic device.

cond-mat.mtrl-sci↗