arXiv subjects

Chi Zhou

Publications and source records attributed to Chi Zhou.

13 recordsLinked to original sources

arXiv2026

Generic Vision and Cross-Attention for Reaction Yield Prediction

Traditional reaction yield prediction is constrained by 1D quantum descriptors that lack explicit spatial information. To address this gap, a dual-modal Vision Cross-Attention architecture is proposed, fusing tabular physical-organic data with 2D molecular topologies. Notably, it is demonstrated that a generic computer vision backbone processing simple 2D skeletal structures independently outperforms purely quantum-based baselines. By synergizing both modalities, superior predictive accuracy compared to traditional methodologies is achieved by the optimal cross-attention framework (Test RMSE = 5.27%). Through mechanistic probing, active, descriptor-guided spatial querying is observed, effectively offloading macroscopic steric identification to the visual pathway. Furthermore, a dynamic chemical hierarchy is learned by the network to heavily prioritize critical steric bottlenecks, such as the aryl halide. Concurrently, residual skip connections are utilized to protect non-spatial electronic parameters from destructive attenuation during fusion. Collectively, a scalable and highly interpretable blueprint is provided for augmenting physical chemistry with deep visual learning.

cs.LG↗

arXiv2026

ChemFusion: A Multimodal Cross-Attention Network for Reaction Yield Prediction

Forecasting the outcomes of transition-metal-catalyzed reactions is notoriously complex due to the interplay of diverse physical and chemical variables. A persistent computational bottleneck has been effectively merging broad electronic descriptors with the localized, three-dimensional geometry of the reactive site. To bridge this representation gap, we present ChemFusion, a hybrid neural network that fuses conventional electronic features with explicit 3D atomic coordinates. Using a cross-attention mechanism, the model enables global electronic states to dynamically attend to specific spatial constraints within un-pooled molecular point clouds. When benchmarked against a diverse library of cross-couplings, this approach delivers exceptional predictive performance, decisively surpassing traditional single-modality frameworks. Importantly, extracting the attention matrices reveals that the architecture autonomously learns to identify and penalize restrictive steric hindrances. This provides a physically grounded interpretability, demonstrating that spatially aware networks can navigate complex reaction sterics that standard statistical models typically miss.

cs.LG↗

arXiv2026

Multimodal Molecular Representation Learning with Graph Neural Networks, Deep & Cross Networks, and SMILES Embeddings

Molecular property prediction often relies on isolated data modalities, where continuous 3D graph neural networks (GNNs) struggle to efficiently capture long-range topological dependencies and exact macroscopic heuristics. In this work, we introduce a parameter-efficient Tri-Branch Modular Fusion Neural Network that synthesizes three orthogonal modalities: 3D spatial geometry (SchNet), discrete topological grammar (SMILES via ChemBERTa), and explicit macroscopic physicochemical descriptors (Deep & Cross Network). By bypassing standard scalar readouts and employing a shared late-fusion architecture, the framework establishes a mathematically rigorous multimodal latent space that effectively resolves the arithmetic and oversmoothing limitations of local message passing. We evaluate the proposed architecture on the QM9 benchmark, targeting the extensive thermodynamic property of atomization energy at 0 K ($U_0^{\mathrm{atom}}$). Through systematic combinatorial ablation and latent bottleneck optimization ($d_e=64$), the tri-modal framework achieves a validation Mean Absolute Error (MAE) of 0.0207 eV. Operating with fewer than one million parameters, this architecture decisively surpasses the sub-chemical accuracy threshold and yields a substantial 20.6% error reduction over a strictly controlled geometric baseline. Ultimately, our findings demonstrate that integrating orthogonal macroscopic and topological data streams provides a synergistic, $\mathcal{O}(1)$ physical shortcut. This multimodal alignment offers a highly efficient alternative to brute-force parameter scaling, establishing a robust surrogate model for high-throughput virtual screening (HTVS) pipelines.

cs.LG↗

arXiv2026

Diffusion-corrected Autoregressive Fourier Neural Operator for Droplet Evolution Prediction

Predicting droplet evolution in material jetting, or Inkjet Printing (IJP), is essential for maintaining printing quality. However, long-horizon forecasts remain challenging due to error accumulation and the complex coupling of process variables. In this work, we introduce the Diffusion-corrected Auto-Regressive Fourier Neural Operator (DiffARFNO), a two-stage framework that combines an autoregressive Fourier-MIONet with a conditional Denoising Diffusion Implicit Model (DDIM) corrector. Fourier-MIONet is trained as a coarse predictor and deployed autoregressively for long-horizon forecasting. In the second stage, a DDIM-based conditional corrector refines the coarse prediction within each sliding window through efficient iterative denoising. By combining coarse predictions from Fourier-MIONet with a DDIM corrector that restores fine details, DiffARFNO aims to provide high-fidelity predictions for long-horizon forecasts. Extensive experiments on droplet datasets from ANSYS Fluent demonstrate that DiffARFNO significantly outperforms existing state-of-the-art models.

cs.LG↗

arXiv2025

Dual Alignment Maximin Optimization for Offline Model-based RL

Offline reinforcement learning agents face significant deployment challenges due to the synthetic-to-real distribution mismatch. While most prior research has focused on improving the fidelity of synthetic sampling and incorporating off-policy mechanisms, the directly integrated paradigm often fails to ensure consistent policy behavior in biased models and underlying environmental dynamics, which inherently arise from discrepancies between behavior and learning policies. In this paper, we first shift the focus from model reliability to policy discrepancies while optimizing for expected returns, and then self-consistently incorporate synthetic data, deriving a novel actor-critic paradigm, Dual Alignment Maximin Optimization (DAMO). It is a unified framework to ensure both model-environment policy consistency and synthetic and offline data compatibility. The inner minimization performs dual conservative value estimation, aligning policies and trajectories to avoid out-of-distribution states and actions, while the outer maximization ensures that policy improvements remain consistent with inner value estimates. Empirical evaluations demonstrate that DAMO effectively ensures model and policy alignments, achieving competitive performance across diverse benchmark tasks.

cs.LG↗

arXiv2024

ASD-Chat: An Innovative Dialogue Intervention System for Children with Autism based on LLM and VB-MAPP

Early diagnosis and professional intervention can help children with autism spectrum disorder (ASD) return to normal life. However, the scarcity and imbalance of professional medical resources currently prevent many autistic children from receiving the necessary diagnosis and intervention. Therefore, numerous paradigms have been proposed that use computer technology to assist or independently conduct ASD interventions, with the aim of alleviating the aforementioned problem. However, these paradigms often lack a foundation in clinical intervention methods and suffer from a lack of personalization. Addressing these concerns, we propose ASD-Chat, a social intervention system based on VB-MAPP (Verbal Behavior Milestones Assessment and Placement Program) and powered by ChatGPT as the backbone for dialogue generation. Specifically, we designed intervention paradigms and prompts based on the clinical intervention method VB-MAPP and utilized ChatGPT's generative capabilities to facilitate social dialogue interventions. Experimental results demonstrate that our proposed system achieves competitive intervention effects to those of professional interventionists, making it a promising tool for long-term interventions in real healthcare scenario in the future.

cs.HC↗

arXiv2024

Mitigating Distribution Shift in Model-based Offline RL via Shifts-aware Reward Learning

Model-based offline reinforcement learning trains policies using pre-collected datasets and learned environment models, eliminating the need for direct real-world environment interaction. However, this paradigm is inherently challenged by distribution shift~(DS). Existing methods address this issue by leveraging off-policy mechanisms and estimating model uncertainty, but they often result in inconsistent objectives and lack a unified theoretical foundation. This paper offers a comprehensive analysis that disentangles the problem into two fundamental components: model bias and policy shift. Our theoretical and empirical investigations reveal how these factors distort value estimation and restrict policy optimization. To tackle these challenges, we derive a novel shifts-aware reward through a unified probabilistic inference framework, which modifies the vanilla reward to refine value learning and facilitate policy training. Building on this, we develop a practical implementation that leverages classifier-based techniques to approximate the adjusted reward for effective policy optimization. Empirical results across multiple benchmarks demonstrate that the proposed approach mitigates distribution shift and achieves superior or comparable performance, validating our theoretical insights.

cs.LG↗

arXiv2024

Know in AdVance: Linear-Complexity Forecasting of Ad Campaign Performance with Evolving User Interest

Real-time Bidding (RTB) advertisers wish to \textit{know in advance} the expected cost and yield of ad campaigns to avoid trial-and-error expenses. However, Campaign Performance Forecasting (CPF), a sequence modeling task involving tens of thousands of ad auctions, poses challenges of evolving user interest, auction representation, and long context, making coarse-grained and static-modeling methods sub-optimal. We propose \textit{AdVance}, a time-aware framework that integrates local auction-level and global campaign-level modeling. User preference and fatigue are disentangled using a time-positioned sequence of clicked items and a concise vector of all displayed items. Cross-attention, conditioned on the fatigue vector, captures the dynamics of user interest toward each candidate ad. Bidders compete with each other, presenting a complete graph similar to the self-attention mechanism. Hence, we employ a Transformer Encoder to compress each auction into embedding by solving auxiliary tasks. These sequential embeddings are then summarized by a conditional state space model (SSM) to comprehend long-range dependencies while maintaining global linear complexity. Considering the irregular time intervals between auctions, we make SSM's parameters dependent on the current auction embedding and the time interval. We further condition SSM's global predictions on the accumulation of local results. Extensive evaluations and ablation studies demonstrate its superiority over state-of-the-art methods. AdVance has been deployed on the Tencent Advertising platform, and A/B tests show a remarkable 4.5\% uplift in Average Revenue per User (ARPU).

cs.IR↗

arXiv2023

Learning-based Two-tiered Online Optimization of Region-wide Datacenter Resource Allocation

Online optimization of resource management for large-scale data centers and infrastructures to meet dynamic capacity reservation demands and various practical constraints (e.g., feasibility and robustness) is a very challenging problem. Mixed Integer Programming (MIP) approaches suffer from recognized limitations in such a dynamic environment, while learning-based approaches may face with prohibitively large state/action spaces. To this end, this paper presents a novel two-tiered online optimization to enable a learning-based Resource Allowance System (RAS). To solve optimal server-to-reservation assignment in RAS in an online fashion, the proposed solution leverages a reinforcement learning (RL) agent to make high-level decisions, e.g., how much resource to select from the Main Switch Boards (MSBs), and then a low-level Mixed Integer Linear Programming (MILP) solver to generate the local server-to-reservation mapping, conditioned on the RL decisions. We take into account fault tolerance, server movement minimization, and network affinity requirements and apply the proposed solution to large-scale RAS problems. To provide interpretability, we further train a decision tree model to explain the learned policies and to prune unreasonable corner cases at the low-level MILP solver, resulting in further performance improvement. Extensive evaluations show that our two-tiered solution outperforms baselines such as pure MIP solver by over $15\%$ while delivering $100\times$ speedup in computation.

cs.NI↗

arXiv2022

Persistence B-Spline Grids: Stable Vector Representation of Persistence Diagrams Based on Data Fitting

Many attempts have been made in recent decades to integrate machine learning (ML) and topological data analysis. A prominent problem in applying persistent homology to ML tasks is finding a vector representation of a persistence diagram (PD), which is a summary diagram for representing topological features. From the perspective of data fitting, a stable vector representation, namely, persistence B-spline grid (PBSG), is proposed based on the efficient technique of progressive-iterative approximation for least-squares B-spline function fitting. We theoretically prove that the PBSG method is stable with respect to the metric of 1-Wasserstein distance defined on the PD space. The proposed method was tested on a synthetic data set, data sets of randomly generated PDs, data of a dynamical system, and 3D CAD models, showing its effectiveness and efficiency

cs.LG↗

arXiv2021

A Predictive Multiphase Model of Silica Aerogels for Building Envelope Insulations

This work develops a multiphase thermomechanical model of porous silica aerogel and implements an uncertainty analysis framework consisting of the Sobol methods for global sensitivity analyses and Bayesian inference using a set of experimental data of silica aerogel. A notable feature of this work is implementing a new noise model within the Bayesian inversion to account for data uncertainty and modeling error. The hyper-parameters in the likelihood balance data misfit and prior contribution to the parameter posteriors and prevent their biased estimation. The results indicate that the uncertainty in solid conductivity and elasticity are the most influential parameters affecting the model output variance. Also, the Bayesian inference shows that despite the microstructural randomness in the thermal measurements, the model captures the data with 2% error. However, the model is inadequate in simulating the stress-strain measurements resulting in significant uncertainty in the computational prediction of a building insulation component.

cond-mat.mtrl-sci↗

arXiv2019

Analysis of RF Energy Harvesting in Uplink-NOMA IoT-based Network

Internet of Things (IoT) systems in general consist of a lot of devices with massive connectivity. Those devices are usually constrained with limited energy supply and can only operate at low power and low rate. In this paper, we investigate a cellular-based IoT system combined with energy harvesting and NOMA. We consider all base stations (BS) and IoT devices follow the Poisson Point Process (PPP) distribution in a given area. The unit time slot is divided into two phases, energy harvesting phase in downlink (DL) and data transmission phase in uplink (uplink). That is, IoT devices will first harvest energy from all BS transmissions and then use the harvested energy to do the NOMA information transmission. We define an energy harvesting circle within which all IoT devices can harvest enough energy for NOMA transmission. The design objective is to maximize the total throughput in uplink within the circle by varying the duration T of energy harvesting phase. In our work, we also consider the inter-cell interference in the throughput calculation. The analysis of Probability Mass Function (PMF) for IoT devices in the energy harvesting circle is also compared with simulation results. It is shown that the BS density needs to be carefully set so that the IoT devices in the energy harvesting circle receive relatively smaller interference and energy circles overlap only with a small probability. Our simulations show that there exists an optimal T to achieve the maximum throughput. When the BSs are densely deployed consequently the total throughput will decrease because of the interference.

eess.SP↗

arXiv2017

Practical Design and Implementation of Metamaterial-Enhanced Magnetic Induction Communication

Although wireless communications in complex environments, such as underground, underwater, and indoor, can enable a large number of novel applications, their performances are constrained by lossy media and complicated structures. Magnetic Induction (MI) has been proved to be an efficient solution to achieve reliable communication in such environments. However, due to the small coil antenna's physical limitation, MI's communication range is still very limited if devices are required to be portable. To this end, Metamaterial-enhanced Magnetic Induction (M$^2$I) communication has been proposed and the theoretical results predict that it can significantly increase the communication performance, namely, data rate and communication range. Nevertheless, currently, the real implementation of M$^2$I is still a challenge and there is no guideline on design and fabrication of spherical metamaterials. In this paper, a practical design is proposed by leveraging a spherical coil array to realize M$^2$I. We prove that the effectively negative permeability can be achieved and there exists a resonance condition where the radiated magnetic field can be significantly amplified. The radiation and communication performances are evaluated and full-wave simulation is conducted to validate the design objectives. By using the spherical coil array-based M$^2$I, the communication range can be significantly extended, exactly as we predicted in the ideal M$^2$I model. Finally, the proposed M$^2$I antenna is implemented and tested in various environments.

physics.class-ph↗