arXiv subjects

Tian Zhao

Publications and source records attributed to Tian Zhao.

At least 19 recordsLinked to original sources

arXiv2026

A Location-Invariant Estimator of Extremal Quantile Treatment Effects for Heavy-Tailed Distributions

Quantile treatment effects (QTEs) measure the effect of a treatment on the distribution of an outcome, and their estimation at extreme quantile levels is of central interest in applications where the target quantiles lie far beyond the range of the data. For heavy-tailed potential outcomes, existing extremal QTE estimators rely on extrapolation combined with a causal extreme value index (EVI) estimator, but the resulting estimator is not invariant under a common location shift of the potential outcome distributions, even though the population QTE is. We address this issue in two steps. First, we adapt the location-invariant Fraga estimator of the EVI to the causal setting using inverse propensity score weighting. Second, we replace the original extrapolation formula with a difference-based scheme, under which the location parameter cancels when quantile differences are taken. The resulting QTE estimator is therefore location invariant. We establish the consistency and asymptotic normality of the proposed extremal QTE estimators, and provide a consistent variance estimator, leading to asymptotically valid inference. A simulation study confirms the location invariance, the stability with respect to the threshold, and the coverage of the proposed methods.

cs.LG↗

arXiv2026

The Essence of Entity Component System

Modern game engines increasingly adopt the Entity Component System (ECS) paradigm as a data-oriented alternative to traditional object-oriented architecture. While ECS promotes modularity and performance through the separation of data and behavior, its practical efficiency depends heavily on the underlying data layout. Despite widespread adoption in frameworks, such as Unity DOTS, Bevy, and Flecs, the semantics of the archetype ECS remain informal and implementation-dependent, limiting rigorous reasoning about determinism, system scheduling, and structural mutations. This work formalizes and experimentally evaluates the archetype ECS. The formal model captures entity creation, component composition, system execution, and archetype migration as compositional state transitions, establishing the core invariants of archetype organization. Using a Tower Defense simulation, we compare the archetype ECS with alternative designs under identical conditions. Results show that the archetype ECS achieves higher frame rate and better frame stability than alternative designs, due to improved cache efficiency and consistent entity access. By uniting formal semantics with empirical validation, this study shows that the archetype ECS outperforms traditional architectures and provides a solid foundation for reasoning about correctness and parallelism.

cs.PL↗

arXiv2026

Evaluating CUDA Tile for AI Workloads on Hopper and Blackwell GPUs

NVIDIA's CUDA Tile (CuTile) introduces a Python-based, tile-centric abstraction for GPU kernel development that aims to simplify programming while retaining Tensor Core and Tensor Memory Accelerator (TMA) efficiency on modern GPUs. We present the first independent, cross-architecture evaluation of CuTile against established approaches such as cuBLAS, Triton, WMMA, and raw SIMT on three NVIDIA GPUs spanning Hopper and Blackwell: H100 NVL, B200, and RTX PRO 6000 Blackwell Server Edition. We benchmark representative AI workloads, including GEMM, fused multi-head attention, and end-to-end LLM inference in BF16/FP16 precision, to assess both performance and portability. Our results show that CuTile effectiveness is strongly workload- and architecture-dependent. On datacenter-class Blackwell (B200), CuTile achieves up to 1007 TFLOP/s for fused attention, outperforming FlashAttention-2 by 2.5x while requiring only 60 lines of Python kernel code. For GEMM, CuTile reaches 52-79% of cuBLAS performance in 22 lines of code (versus 123 for WMMA), making it a practical replacement for hand-written CUDA kernels but not yet for vendor-optimized libraries. However, the same CuTile attention kernel achieves only 53% of FlashAttention-2 throughput on RTX PRO 6000 (sm_120), exposing significant cross-architecture optimization gaps. In contrast, Triton sustains 62-101% of cuBLAS performance across all tested platforms without architecture-specific tuning, demonstrating substantially stronger portability.

cs.LG↗

arXiv2026

Hybrid JIT-CUDA Graph Optimization for Low-Latency Large Language Model Inference

Large Language Models (LLMs) have achieved strong performance across natural language and multimodal tasks, yet their practical deployment remains constrained by inference latency and kernel launch overhead, particularly in interactive, short-sequence settings. This paper presents a hybrid runtime framework that combines Just-In-Time (JIT) compilation with CUDA Graph execution to reduce launch overhead while preserving runtime flexibility during autoregressive decoding. The framework partitions transformer inference into static components executed via CUDA Graph replay and dynamic components handled through JIT-compiled kernels, enabling asynchronous graph capture and reuse across decoding steps. We evaluate the proposed approach on LLaMA-2 7B using single-GPU, batch-size-one inference across prompt lengths from 10 to 500 tokens. Experimental results show that the hybrid runtime reduces Time-to-First-Token (TTFT) by up to 66.0% and achieves lower P99 latency compared with TensorRT-LLM in this regime. These results indicate that hybrid JIT-CUDA Graph execution can effectively reduce inference latency and variance for short-sequence LLM workloads, making it a practical optimization strategy for latency-sensitive AI applications.

cs.LG↗

arXiv2024

Medical Multimodal Foundation Models in Clinical Diagnosis and Treatment: Applications, Challenges, and Future Directions

Recent advancements in deep learning have significantly revolutionized the field of clinical diagnosis and treatment, offering novel approaches to improve diagnostic precision and treatment efficacy across diverse clinical domains, thus driving the pursuit of precision medicine. The growing availability of multi-organ and multimodal datasets has accelerated the development of large-scale Medical Multimodal Foundation Models (MMFMs). These models, known for their strong generalization capabilities and rich representational power, are increasingly being adapted to address a wide range of clinical tasks, from early diagnosis to personalized treatment strategies. This review offers a comprehensive analysis of recent developments in MMFMs, focusing on three key aspects: datasets, model architectures, and clinical applications. We also explore the challenges and opportunities in optimizing multimodal representations and discuss how these advancements are shaping the future of healthcare by enabling improved patient outcomes and more efficient clinical workflows.

cs.AI↗

arXiv2024

GCtx-UNet: Efficient Network for Medical Image Segmentation

Medical image segmentation is crucial for disease diagnosis and monitoring. Though effective, the current segmentation networks such as UNet struggle with capturing long-range features. More accurate models such as TransUNet, Swin-UNet, and CS-UNet have higher computation complexity. To address this problem, we propose GCtx-UNet, a lightweight segmentation architecture that can capture global and local image features with accuracy better or comparable to the state-of-the-art approaches. GCtx-UNet uses vision transformer that leverages global context self-attention modules joined with local self-attention to model long and short range spatial dependencies. GCtx-UNet is evaluated on the Synapse multi-organ abdominal CT dataset, the ACDC cardiac MRI dataset, and several polyp segmentation datasets. In terms of Dice Similarity Coefficient (DSC) and Hausdorff Distance (HD) metrics, GCtx-UNet outperformed CNN-based and Transformer-based approaches, with notable gains in the segmentation of complex and small anatomical structures. Moreover, GCtx-UNet is much more efficient than the state-of-the-art approaches with smaller model size, lower computation workload, and faster training and inference speed, making it a practical choice for clinical applications.

eess.IV↗

arXiv2024

JungleGPT: Designing and Optimizing Compound AI Systems for E-Commerce

LLMs have significantly advanced the e-commerce industry by powering applications such as personalized recommendations and customer service. However, most current efforts focus solely on monolithic LLMs and fall short in addressing the complexity and scale of real-world e-commerce scenarios. In this work, we present JungleGPT, the first compound AI system tailored for real-world e-commerce applications. We outline the system's design and the techniques used to optimize its performance for practical use cases, which have proven to reduce inference costs to less than 1% of what they would be with a powerful, monolithic LLM.

cs.IR↗

arXiv2024

Extended Kalman Filtering for Recursive Online Discrete-Time Inverse Optimal Control

We formulate the discrete-time inverse optimal control problem of inferring unknown parameters in the objective function of an optimal control problem from measurements of optimal states and controls as a nonlinear filtering problem. This formulation enables us to propose a novel extended Kalman filter (EKF) for solving inverse optimal control problems in a computationally efficient recursive online manner that requires only a single pass through the measurement data. Importantly, we show that the Jacobians required to implement our EKF can be computed efficiently by exploiting recent Pontryagin differentiable programming results, and that our consideration of an EKF enables the development of first-of-their-kind theoretical error guarantees for online inverse optimal control with noisy incomplete measurements. Our proposed EKF is shown to be significantly faster than an alternative unscented Kalman filter-based approach.

eess.SY↗

arXiv2023

Transfer Learning for Microstructure Segmentation with CS-UNet: A Hybrid Algorithm with Transformer and CNN Encoders

Transfer learning improves the performance of deep learning models by initializing them with parameters pre-trained on larger datasets. Intuitively, transfer learning is more effective when pre-training is on the in-domain datasets. A recent study by NASA has demonstrated that the microstructure segmentation with encoder-decoder algorithms benefits more from CNN encoders pre-trained on microscopy images than from those pre-trained on natural images. However, CNN models only capture the local spatial relations in images. In recent years, attention networks such as Transformers are increasingly used in image analysis to capture the long-range relations between pixels. In this study, we compare the segmentation performance of Transformer and CNN models pre-trained on microscopy images with those pre-trained on natural images. Our result partially confirms the NASA study that the segmentation performance of out-of-distribution images (taken under different imaging and sample conditions) is significantly improved when pre-training on microscopy images. However, the performance gain for one-shot and few-shot learning is more modest with Transformers. We also find that for image segmentation, the combination of pre-trained Transformers and CNN encoders are consistently better than pre-trained CNN encoders alone. Our dataset (of about 50,000 images) combines the public portion of the NASA dataset with additional images we collected. Even with much less training data, our pre-trained models have significantly better performance for image segmentation. This result suggests that Transformers and CNN complement each other and when pre-trained on microscopy images, they are more beneficial to the downstream tasks.

cs.CV↗

arXiv2022

Stardust: Compiling Sparse Tensor Algebra to a Reconfigurable Dataflow Architecture

We introduce Stardust, a compiler that compiles sparse tensor algebra to reconfigurable dataflow architectures (RDAs). Stardust introduces new user-provided data representation and scheduling language constructs for mapping to resource-constrained accelerated architectures. Stardust uses the information provided by these constructs to determine on-chip memory placement and to lower to the Capstan RDA through a parallel-patterns rewrite system that targets the Spatial programming model. The Stardust compiler is implemented as a new compilation path inside the TACO open-source system. Using cycle-accurate simulation, we demonstrate that Stardust can generate more Capstan tensor operations than its authors had implemented and that it results in 138$\times$ better performance than generated CPU kernels and 41$\times$ better performance than generated GPU kernels.

cs.PL↗

arXiv2022

Ultrasensitive atomic comagnetometer with enhanced nuclear spin coherence

Achieving high energy resolution in spin systems is important for fundamental physics research and precision measurements, with alkali-noble-gas comagnetometers being among the best available sensors. We found a new relaxation mechanism in such devices, the gradient of the Fermi-contact-interaction field that dominates the relaxation of hyperpolarized nuclear spins. We report on precise control over spin distribution, demonstrating a tenfold increase of nuclear spin hyperpolarization and transverse coherence time with optimal hybrid optical pumping. Operating in the self-compensation regime, our $^{21}$Ne-Rb-K comagnetometer achieves an ultrahigh inertial rotation sensitivity of $3\times10^{-8}$\,rad/s/Hz$^{1/2}$ in the frequency range from 0.2 to 1.0 Hz, which is equivalent to the energy resolution of $3.1\times 10^{-23}$\,eV/Hz$^{1/2}$. We propose to use this comagnetometer to search for exotic spin-dependent interactions involving proton and neutron spins. The projected sensitivity surpasses the previous experimental and astrophysical limits by more than four orders of magnitude.

physics.atom-ph↗

arXiv2022

Computer Vision Methods for the Microstructural Analysis of Materials: The State-of-the-art and Future Perspectives

Finding quantitative descriptors representing the microstructural features of a given material is an ongoing research area in the paradigm of Materials-by-Design. Historically, microstructural analysis mostly relies on qualitative descriptions. However, to build a robust and accurate process-structure-properties relationship, which is required for designing new advanced high-performance materials, the extraction of quantitative and meaningful statistical data from the microstructural analysis is a critical step. In recent years, computer vision (CV) methods, especially those which are centered around convolutional neural network (CNN) algorithms have shown promising results for this purpose. This review paper focuses on the state-of-the-art CNN-based techniques that have been applied to various multi-scale microstructural image analysis tasks, including classification, object detection, segmentation, feature extraction, and reconstruction. Additionally, we identified the main challenges with regard to the application of these methods to materials science research. Finally, we discussed some possible future directions of research in this area. In particular, we emphasized the application of transformer-based models and their capabilities to improve the microstructural analysis of materials.

cond-mat.mtrl-sci↗

arXiv2022

Efficient Memory Partitioning in Software Defined Hardware

As programmers turn to software-defined hardware (SDH) to maintain a high level of productivity while programming hardware to run complex algorithms, heavy-lifting must be done by the compiler to automatically partition on-chip arrays. In this paper, we introduce an automatic memory partitioning system that can quickly compute more efficient partitioning schemes than prior systems. Our system employs a variety of resource-saving optimizations and an ML cost model to select the best partitioning scheme from an array of candidates. We compared our system against various state-of-the-art SDH compilers and FPGAs on a variety of benchmarks and found that our system generates solutions that, on average, consume 40.3% fewer logic resources, 78.3% fewer FFs, 54.9% fewer Block RAMs (BRAMs), and 100% fewer DSPs.

cs.AR↗

arXiv2021

Capstan: A Vector RDA for Sparsity

This paper proposes Capstan: a scalable, parallel-patterns-based, reconfigurable dataflow accelerator (RDA) for sparse and dense tensor applications. Instead of designing for one application, we start with common sparse data formats, each of which supports multiple applications. Using a declarative programming model, Capstan supports application-independent sparse iteration and memory primitives that can be mapped to vectorized, high-performance hardware. We optimize random-access sparse memories with configurable out-of-order execution to increase SRAM random-access throughput from 32% to 80%. For a variety of sparse applications, Capstan with DDR4 memory is 18x faster than a multi-core CPU baseline, while Capstan with HBM2 memory is 16x faster than an Nvidia V100 GPU. For sparse applications that can be mapped to Plasticine, a recent dense RDA, Capstan is 7.6x to 365x faster and only 16% larger.

cs.AR↗

arXiv2020

Unexpected Near-Infrared to Visible Non-linear Optical Properties from Two-Dimensional Polar Metals

Near-infrared-to-visible second harmonic generation from air-stable two-dimensional polar gallium and indium metals is described. The photonic properties of 2D metals - including the largest second-order susceptibilities reported for metals (approaching 10nm$^2$/V) - are determined by the atomic-level structure and bonding of two-to-three-atom-thick crystalline films. The bond character evolved from covalent to metallic over a few atomic layers, changing the out-of-plane metal-metal bond distances by approximately ten percent (0.2 $\unicode{x212B}$), resulting in symmetry breaking and an axial electrostatic dipole that mediated the large nonlinear response. Two different orientations of the crystalline metal atoms, corresponding to lateral displacements < 2 $\unicode{x212B}$, persisted in separate micron-scale terraces to generate distinct harmonic polarizations. This strong atomic-level structure-property interplay suggests metal photonic properties can be controlled with atomic precision.

cond-mat.mtrl-sci↗

arXiv2019

Analysis of DAWNBench, a Time-to-Accuracy Machine Learning Performance Benchmark

Researchers have proposed hardware, software, and algorithmic optimizations to improve the computational performance of deep learning. While some of these optimizations perform the same operations faster (e.g., increasing GPU clock speed), many others modify the semantics of the training procedure (e.g., reduced precision), and can impact the final model's accuracy on unseen data. Due to a lack of standard evaluation criteria that considers these trade-offs, it is difficult to directly compare these optimizations. To address this problem, we recently introduced DAWNBench, a benchmark competition focused on end-to-end training time to achieve near-state-of-the-art accuracy on an unseen dataset---a combined metric called time-to-accuracy (TTA). In this work, we analyze the entries from DAWNBench, which received optimized submissions from multiple industrial groups, to investigate the behavior of TTA as a metric as well as trends in the best-performing entries. We show that TTA has a low coefficient of variation and that models optimized for TTA generalize nearly as well as those trained using standard methods. Additionally, even though DAWNBench entries were able to train ImageNet models in under 3 minutes, we find they still underutilize hardware capabilities such as Tensor Cores. Furthermore, we find that distributed entries can spend more than half of their time on communication. We show similar findings with entries to the MLPERF v0.5 benchmark.

cs.LG↗

arXiv2019

Serving Recurrent Neural Networks Efficiently with a Spatial Accelerator

Recurrent Neural Network (RNN) applications form a major class of AI-powered, low-latency data center workloads. Most execution models for RNN acceleration break computation graphs into BLAS kernels, which lead to significant inter-kernel data movement and resource underutilization. We show that by supporting more general loop constructs that capture design parameters in accelerators, it is possible to improve resource utilization using cross-kernel optimization without sacrificing programmability. Such abstraction level enables a design space search that can lead to efficient usage of on-chip resources on a spatial architecture across a range of problem sizes. We evaluate our optimization strategy on such abstraction with DeepBench using a configurable spatial accelerator. We demonstrate that this implementation provides a geometric speedup of 30x in performance, 1.6x in area, and 2x in power efficiency compared to a Tesla V100 GPU, and a geometric speedup of 2x compared to Microsoft Brainwave implementation on a Stratix 10 FPGA.

cs.DC↗

arXiv2019

Epitaxial Graphene Intercalation: A Route to Graphene Modulation and Unique 2D Materials

Intercalation of atomic species through epitaxial graphene layers began only a few years following its initial report in 2004. The impact of intercalation on the electronic properties of the graphene is well known; however, the intercalant itself can also exhibit intriguing properties not found in nature. This suggests that a shift in the focus of epitaxial graphene intercalation studies may lead to fruitful exploration of many new forms of traditionally 3D materials. In the following forward-looking review, we summarize the primary techniques used to achieve and characterize EG intercalation, and introduce a new, facile approach to readily achieve metal intercalation at the graphene/silicon carbide interface. We show that simple thermal evaporation-based methods can effectively replace complicated synthesis techniques to realize large-scale intercalation of non-refractory metals. We also show that these methods can be extended to the formation of compound materials based on intercalation. Two-dimensional (2D) silver (2D-Ag) and large-scale 2D gallium nitride (2D-GaNx) are used to demonstrate these approaches.

cond-mat.mtrl-sci↗