Author name cluster

Xiaoyu Jiang

Possible papers associated with this exact author name in Arrow. This page groups case-insensitive exact name matches and is not a full identity disambiguation profile.

5 papers

2 author rows

ICML Conference 2025 Conference Paper

Neighbour-Driven Gaussian Process Variational Autoencoders for Scalable Structured Latent Modelling

Xinxing Shi
Xiaoyu Jiang
Mauricio A. Álvarez

Gaussian Process (GP) Variational Autoencoders (VAEs) extend standard VAEs by replacing the fully factorised Gaussian prior with a GP prior, thereby capturing richer correlations among latent variables. However, performing exact GP inference in large-scale GPVAEs is computationally prohibitive, often forcing existing approaches to rely on restrictive kernel assumptions or large sets of inducing points. In this work, we propose a neighbour-driven approximation strategy that exploits local adjacencies in the latent space to achieve scalable GPVAE inference. By confining computations to the nearest neighbours of each data point, our method preserves essential latent dependencies, allowing more flexible kernel choices and mitigating the need for numerous inducing points. Through extensive experiments on tasks including representation learning, data imputation, and conditional generation, we demonstrate that our approach outperforms other GPVAE variants in both predictive performance and computational efficiency.

Details

TMLR Journal 2025 Journal Article

Scalable Multi-Output Gaussian Processes with Stochastic Variational Inference

Xiaoyu Jiang
Sokratia Georgaka
Magnus Rattray
Mauricio A Álvarez

The Multi-Output Gaussian Process (MOGP) is a popular tool for modelling data from multiple sources. A typical choice to build a covariance function for a MOGP is the Linear Model of Coregionalisation (LMC) which parametrically models the covariance between outputs. The Latent Variable MOGP (LV-MOGP) generalises this idea by modelling the covariance between outputs using a kernel applied to latent variables, one per output, leading to a flexible MOGP model that allows efficient generalisation to new outputs with few data points. The computational complexity in LV-MOGP grows linearly with the number of outputs, which makes it unsuitable for problems with a large number of outputs. In this paper, we propose a stochastic variational inference approach for the LV-MOGP that allows mini-batches for both inputs and outputs, making computational complexity per training iteration independent of the number of outputs. We demonstrate the performance of the model by benchmarking against some other MOGP models in several real-world datasets, including spatial-temporal climate modelling and spatial transcriptomics.

PDF Details

EAAI Journal 2024 Journal Article

Chronicle knowledge-based multi-level response prediction for predictive control by forest models in process industry

Linjin Sun
Yangjian Ji
Zheren Zhu
Xiaoyu Jiang
Xiaoyang Zhu
Nian Zhang

The control output response prediction is confirmed to be crucial for predictive control in process industries. The data-driven approaches that adopt system-independent predictors for output response predictions of multiple controlled variables are promising due to their easy implementation using available open-source models. However, the intrinsic knowledge among controlled variables is ignored, for it cannot be fed into the data-driven models directly, including control events, control antecedents, and temporal constraint information, which inevitably limits the predictive performance. This study proposes a hybrid control output response prediction method embedded with chronicle knowledge and a data-driven model for predictive control. It starts from the knowledge discovery and instance pool establishment by symbolic techniques. Control intrinsic knowledge extracted from the instance pool will be integrated into the modeling of forest model-based predictors, where the knowledge will be leveraged for predictor structure designing and temporal constraints estimation. In this way, tailored predictors with multi-level structures will be devised to capture the intrinsic correlation of controlled variables with different working levels. The control output response predictors will finally be introduced into the design of predictive controllers to make multi-output response predictions. The experimental results show the superiority of the proposed method over baselines in the prediction accuracy as well as the capability of set-point tracking and disturbance rejection.

Details DOI

NeurIPS Conference 2024 Conference Paper

Rethinking the Diffusion Models for Missing Data Imputation: A Gradient Flow Perspective

Zhichao Chen
Haoxuan Li
Fangyikang Wang
Odin Zhang
Hu Xu
Xiaoyu Jiang
Zhihuan Song
Hao Wang

Diffusion models have demonstrated competitive performance in missing data imputation (MDI) task. However, directly applying diffusion models to MDI produces suboptimal performance due to two primary defects. First, the sample diversity promoted by diffusion models hinders the accurate inference of missing values. Second, data masking reduces observable indices for model training, obstructing imputation performance. To address these challenges, we introduce $\underline{\text{N}}$egative $\underline{\text{E}}$ntropy-regularized $\underline{\text{W}}$asserstein gradient flow for $\underline{\text{Imp}}$utation (NewImp), enhancing diffusion models for MDI from a gradient flow perspective. To handle the first defect, we incorporate a negative entropy regularization term into the cost functional to suppress diversity and improve accuracy. To handle the second defect, we demonstrate that the imputation procedure of NewImp, induced by the conditional distribution-related cost functional, can equivalently be replaced by that induced by the joint distribution, thereby naturally eliminating the need for data masking. Extensive experiments validate the effectiveness of our method. Code is available at [https: //github. com/JustusvLiebig/NewImp](https: //github. com/JustusvLiebig/NewImp).

PDF Details DOI

YNIMG Journal 2014 Journal Article

Mapping mean axon diameter and axonal volume fraction by MRI using temporal diffusion spectroscopy

Junzhong Xu
Hua Li
Kevin D. Harkins
Xiaoyu Jiang
Jingping Xie
Hakmook Kang
Mark D. Does
John C. Gore

Mapping mean axon diameter and intra-axonal volume fraction may have significant clinical potential because nerve conduction velocity is directly dependent on axon diameter, and several neurodegenerative diseases affect axons of specific sizes and alter axon counts. Diffusion-weighted MRI methods based on the pulsed gradient spin echo (PGSE) sequence have been reported to be able to assess axon diameter and volume fraction non-invasively. However, due to the relatively long diffusion times used, e. g. >20ms, the sensitivity to small axons (diameter<2μm) is low, and the derived mean axon diameter has been reported to be overestimated. In the current study, oscillating gradient spin echo (OGSE) diffusion sequences with variable frequency gradients were used to assess rat spinal white matter tracts with relatively short effective diffusion times (1–5ms). In contrast to previous PGSE-based methods, the extra-axonal diffusion cannot be modeled as hindered (Gaussian) diffusion when short diffusion times are used. Appropriate frequency-dependent rates are therefore incorporated into our analysis and validated by histology-based computer simulation of water diffusion. OGSE data were analyzed to derive mean axon diameters and intra-axonal volume fractions of rat spinal white matter tracts (mean axon diameter of ~1. 27–5. 54μm). The estimated values were in good agreement with histology, including the small axon diameters (<2. 5μm). This study establishes a framework for the quantification of nerve morphology using the OGSE method with high sensitivity to small axons.

Details DOI