Author name cluster

Jiaqi Gu

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.

9 papers

1 author row

JMLR Journal 2025 Journal Article

Hierarchical and Stochastic Crystallization Learning: Geometrically Leveraged Nonparametric Regression with Delaunay Triangulation

Jiaqi Gu
Guosheng Yin

High-dimensionality is known to be the bottleneck for both nonparametric regression and the Delaunay triangulation. To efficiently exploit the advantage of the Delaunay triangulation in utilizing geometry information for nonparametric regression without conducting the Delaunay triangulation for the entire feature space, we develop the crystallization search for the neighbor Delaunay simplices of the target point similar to crystal growth and estimate the conditional expectation function by fitting a local linear model to the data points of the constructed Delaunay simplices. Because the shapes and volumes of Delaunay simplices are adaptive to the density of feature data points, our method selects neighbor data points more uniformly in all directions in comparison with Euclidean distance based methods and thus it is more robust to the local geometric structure of the data. We further develop the stochastic approach to hyperparameter selection and the hierarchical crystallization learning under multimodal feature data densities, where an approximate global Delaunay triangulation is obtained by first triangulating the local centers and then constructing local Delaunay triangulations in parallel. We study the asymptotic properties of our method and conduct numerical experiments on both synthetic and real data to demonstrate the advantages of our method over the existing ones. [abs] [ pdf ][ bib ] &copy JMLR 2025. ( edit, beta )

PDF Details

NeurIPS Conference 2025 Conference Paper

Knowledge-based Visual Question Answer with Multimodal Processing, Retrieval and Filtering

yuyang Hong
Jiaqi Gu
Yang Qi
Lubin Fan
Yue Wu
Ying Wang
Kun Ding
Shiming Xiang

The task of Knowlegde-Based Visual Question Answering (KB-VQA) requires the model to understand visual features and retrieve external knowledge. Retrieval-Augmented Generation (RAG) have been employed to address this problem through knowledge base querying. However, existing work demonstrate two limitations: insufficient interactivity during knowledge retrieval and ineffective organization of retrieved information for Visual-Language Model (VLM). To address these challenges, we propose a three-stage visual language model with Process, Retrieve and Filter (VLM-PRF) framework. For interactive retrieval, VLM-PRF uses reinforcement learning (RL) to guide the model to strategically process information via tool-driven operations. For knowledge filtering, our method trains the VLM to transform the raw retrieved information into into task-specific knowledge. With a dual reward as supervisory signals, VLM-PRF successfully enable model to optimize retrieval strategies and answer generation capabilities simultaneously. Experiments on two datasets demonstrate the effectiveness of our framework.

PDF Details

NeurIPS Conference 2024 Conference Paper

PACE: Pacing Operator Learning to Accurate Optical Field Simulation for Complicated Photonic Devices

Hanqing Zhu
Wenyan Cong
Guojin Chen
Shupeng Ning
Ray T. Chen
Jiaqi Gu
David Z. Pan

Electromagnetic field simulation is central to designing, optimizing, and validating photonic devices and circuits. However, costly computation associated with numerical simulation poses a significant bottleneck, hindering scalability and turnaround time in the photonic circuit design process. Neural operators offer a promising alternative, but existing SOTA approaches, Neurolight, struggle with predicting high-fidelity fields for real-world complicated photonic devices, with the best reported 0. 38 normalized mean absolute error in Neurolight. The interplays of highly complex light-matter interaction, e. g. , scattering and resonance, sensitivity to local structure details, non-uniform learning complexity for full-domain simulation, and rich frequency information, contribute to the failure of existing neural PDE solvers. In this work, we boost the prediction fidelity to an unprecedented level for simulating complex photonic devices with a novel operator design driven by the above challenges. We propose a novel cross-axis factorized PACE operator with a strong long-distance modeling capacity to connect the full-domain complex field pattern with local device structures. Inspired by human learning, we further divide and conquer the simulation task for extremely hard cases into two progressively easy tasks, with a first-stage model learning an initial solution refined by a second model. On various complicated photonic device benchmarks, we demonstrate one sole PACE model is capable of achieving 73% lower error with 50% fewer parameters compared with various recent ML for PDE solvers. The two-stage setup further advances high-fidelity simulation for even more intricate cases. In terms of runtime, PACE demonstrates 154-577x and 11. 8-12x simulation speedup over numerical solver using scipy or highly-optimized pardiso solver, respectively. We open-sourced the code and complicated optical device dataset at PACE-Light.

PDF Details DOI

NeurIPS Conference 2023 Conference Paper

Pre-RMSNorm and Pre-CRMSNorm Transformers: Equivalent and Efficient Pre-LN Transformers

Zixuan Jiang
Jiaqi Gu
Hanqing Zhu
David Pan

Transformers have achieved great success in machine learning applications. Normalization techniques, such as Layer Normalization (LayerNorm, LN) and Root Mean Square Normalization (RMSNorm), play a critical role in accelerating and stabilizing the training of Transformers. While LayerNorm recenters and rescales input vectors, RMSNorm only rescales the vectors by their RMS value. Despite being more computationally efficient, RMSNorm may compromise the representation ability of Transformers. There is currently no consensus regarding the preferred normalization technique, as some models employ LayerNorm while others utilize RMSNorm, especially in recent large language models. It is challenging to convert Transformers with one normalization to the other type. While there is an ongoing disagreement between the two normalization types, we propose a solution to unify two mainstream Transformer architectures, Pre-LN and Pre-RMSNorm Transformers. By removing the inherent redundant mean information in the main branch of Pre-LN Transformers, we can reduce LayerNorm to RMSNorm, achieving higher efficiency. We further propose the Compressed RMSNorm (CRMSNorm) and Pre-CRMSNorm Transformer based on a lossless compression of the zero-mean vectors. We formally establish the equivalence of Pre-LN, Pre-RMSNorm, and Pre-CRMSNorm Transformer variants in both training and inference. It implies that Pre-LN Transformers can be substituted with Pre-(C)RMSNorm counterparts at almost no cost, offering the same arithmetic functionality along with free efficiency improvement. Experiments demonstrate that we can reduce the training and inference time of Pre-LN Transformers by 1% - 10%.

PDF Details

NeurIPS Conference 2022 Conference Paper

NeurOLight: A Physics-Agnostic Neural Operator Enabling Parametric Photonic Device Simulation

Jiaqi Gu
Zhengqi Gao
Chenghao Feng
Hanqing Zhu
Ray Chen
Duane Boning
David Pan

Optical computing has become emerging technology in next-generation efficient artificial intelligence (AI) due to its ultra-high speed and efficiency. Electromagnetic field simulation is critical to the design, optimization, and validation of photonic devices and circuits. However, costly numerical simulation significantly hinders the scalability and turn-around time in the photonic circuit design loop. Recently, physics-informed neural networks were proposed to predict the optical field solution of a single instance of a partial differential equation (PDE) with predefined parameters. Their complicated PDE formulation and lack of efficient parametrization mechanism limit their flexibility and generalization in practical simulation scenarios. In this work, for the first time, a physics-agnostic neural operator-based framework, dubbed NeurOLight, is proposed to learn a family of frequency-domain Maxwell PDEs for ultra-fast parametric photonic device simulation. Specifically, we discretize different devices into a unified domain, represent parametric PDEs with a compact wave prior, and encode the incident light via masked source modeling. We design our model to have parameter-efficient cross-shaped NeurOLight blocks and adopt superposition-based augmentation for data-efficient learning. With those synergistic approaches, NeurOLight demonstrates 2-orders-of-magnitude faster simulation speed than numerical solvers and outperforms prior NN-based models by ~54% lower prediction error using ~44% fewer parameters.

PDF Details

AAAI Conference 2021 Conference Paper

Efficient On-Chip Learning for Optical Neural Networks Through Power-Aware Sparse Zeroth-Order Optimization

Jiaqi Gu
Chenghao Feng
Zheng Zhao
Zhoufeng Ying
Ray T. Chen
David Z. Pan

Optical neural networks (ONNs) have demonstrated recordbreaking potential in high-performance neuromorphic computing due to their ultra-high execution speed and low energy consumption. However, current learning protocols fail to provide scalable and efficient solutions to photonic circuit optimization in practical applications. In this work, we propose a novel on-chip learning framework to release the full potential of ONNs for power-efficient in situ training. Instead of deploying implementation-costly back-propagation, we directly optimize the device configurations with computation budgets and power constraints. We are the first to model the ONN on-chip learning as a resource-constrained stochastic noisy zeroth-order optimization problem, and propose a novel mixed-training strategy with two-level sparsity and poweraware dynamic pruning to offer a scalable on-chip training solution in practical ONN deployment. Compared with previous methods, we are the first to optimize over 2, 500 optical components on chip. We can achieve much better optimization stability, 3. 7×-7. 6× higher efficiency, and save >90% power under practical device variations and thermal crosstalk.

PDF Details

NeurIPS Conference 2021 Conference Paper

L2ight: Enabling On-Chip Learning for Optical Neural Networks via Efficient in-situ Subspace Optimization

Jiaqi Gu
Hanqing Zhu
Chenghao Feng
Zixuan Jiang
Ray Chen
David Pan

Silicon-photonics-based optical neural network (ONN) is a promising hardware platform that could represent a paradigm shift in efficient AI with its CMOS-compatibility, flexibility, ultra-low execution latency, and high energy efficiency. In-situ training on the online programmable photonic chips is appealing but still encounters challenging issues in on-chip implementability, scalability, and efficiency. In this work, we propose a closed-loop ONN on-chip learning framework L2ight to enable scalable ONN mapping and efficient in-situ learning. L2ight adopts a three-stage learning flow that first calibrates the complicated photonic circuit states under challenging physical constraints, then performs photonic core mapping via combined analytical solving and zeroth-order optimization. A subspace learning procedure with multi-level sparsity is integrated into L2ight to enable in-situ gradient evaluation and fast adaptation, unleashing the power of optics for real on-chip intelligence. Extensive experiments demonstrate our proposed L2ight outperforms prior ONN training protocols with 3-order-of-magnitude higher scalability and over 30x better efficiency, when benchmarked on various models and learning tasks. This synergistic framework is the first scalable on-chip learning solution that pushes this emerging field from intractable to scalable and further to efficient for next-generation self-learnable photonic neural chips. From a co-design perspective, L2ight also provides essential insights for hardware-restricted unitary subspace optimization and efficient sparse training. We open-source our framework at the link.

PDF Details

EAAI Journal 2021 Journal Article

PEAVC: An improved minimum vertex cover solver for massive sparse graphs

Jiaqi Gu
Ping Guo

Several important applications related to complex network analysis require finding small vertex covers in massive sparse graphs. To fulfill this task, this paper proposes a general algorithm framework named PEAF, which includes preprocessing stage, solving stage, and inverse-processing stage. Based on PEAF, a minimum vertex cover (MinVC) solver PEAVC is developed, which uses PreP to reduce the graph, BGVC to obtain a vertex cover of bipartite-graph components, FastVC2 to solve the connected components left, and Inv_PreP to get a vertex cover of the original problem. Computational experiments on 90 massive REAL-WORLD benchmark graphs indicate that PreP can reduce the vertex number by 83. 25% on average, which is superior to other graph reduction methods. PEAVC performs remarkably well by discovering 5 best-known results (new upper bounds) never reported in the literature, match the best known results for 63 other instances and obtain exact MinVCs for 55 instances. Experiments also show that PEAVC has an extremely high performance.

Details DOI

I&C Journal 2018 Journal Article

User-friendly temporal queries on historical knowledge bases

Carlo Zaniolo
Shi Gao
Maurizio Atzori
Muhao Chen
Jiaqi Gu

DBpedia and other RFD-encoded Knowledge Bases (KB)s give users access to encyclopedic knowledge via SPARQL queries. As the world evolves, the KBs are updated, and the history of entities and their properties becomes of great interest. Thus, we need powerful tools and friendly interfaces to query histories and flash-back to the past. Here, we propose (i) a point-based temporal extension of SPARQL, called SPARQLT, which enables simple and concise expression of temporal queries, and (ii) an extension of Wikipedia Infoboxes to support user-friendly by-example temporal queries implemented by mapping them into SPARQLT. Our main-memory RDF-TX system supports such queries efficiently using Multi-Version B+ trees, compressed indexes, and query optimization techniques, which achieve performance and scalability, as demonstrated by experiments on historical datasets including Cliopedia derived from Wikipedia dumps. We finally discuss how provenance information can be used to add valid-time features to these transaction-time KBs.

Details DOI