Jun Han

Qitong Sun, Jun Han, Tianlin Li et al.

Improving GPU kernel efficiency is crucial for advancing AI systems. Recent work has explored leveraging large language models (LLMs) for GPU kernel generation and optimization. However, existing LLM-based kernel optimization pipelines typically rely on opaque, implicitly learned heuristics within the LLMs to determine optimization strategies. This leads to inefficient trial-and-error and weakly interpretable optimizations. Our key insight is to replace implicit heuristics with expert optimization skills that are knowledge-driven and aware of task trajectories. Specifically, we present KernelSkill, a multi-agent framework with a dual-level memory architecture. KernelSkill operates by coordinating agents with long-term memory of reusable expert skills and short-term memory to prevent repetitive backtracking. On KernelBench Levels 1-3, KernelSkill achieves a 100% success rate and average speedups of 5.44x, 2.82x, and 1.92x over Torch Eager on Levels 1, 2, and 3, respectively, outperforming prior baselines. Code is available at https://github.com/0satan0/KernelMem/.

Chaoli Wang, Jun Han

Since 2016, we have witnessed the tremendous growth of artificial intelligence+visualization (AI+VIS) research. However, existing survey papers on AI+VIS focus on visual analytics and information visualization, not scientific visualization (SciVis). In this paper, we survey related deep learning (DL) works in SciVis, specifically in the direction of DL4SciVis: designing DL solutions for solving SciVis problems. To stay focused, we primarily consider works that handle scalar and vector field data but exclude mesh data. We classify and discuss these works along six dimensions: domain setting, research task, learning type, network architecture, loss function, and evaluation metric. The paper concludes with a discussion of the remaining gaps to fill along the discussed dimensions and the grand challenges we need to tackle as a community. This state-of-the-art survey guides SciVis researchers in gaining an overview of this emerging topic and points out future directions to grow this research.

Zhaoqiang Liu, Jiulong Liu, Subhroshekhar Ghosh et al.

In this paper, we study the problem of principal component analysis with generative modeling assumptions, adopting a general model for the observed matrix that encompasses notable special cases, including spiked matrix recovery and phase retrieval. The key assumption is that the underlying signal lies near the range of an $L$-Lipschitz continuous generative model with bounded $k$-dimensional inputs. We propose a quadratic estimator, and show that it enjoys a statistical rate of order $\sqrt{\frac{k\log L}{m}}$, where $m$ is the number of samples. We also provide a near-matching algorithm-independent lower bound. Moreover, we provide a variant of the classic power method, which projects the calculated data onto the range of the generative model during each iteration. We show that under suitable conditions, this method converges exponentially fast to a point achieving the above-mentioned statistical rate. We perform experiments on various image datasets for spiked matrix and phase retrieval models, and illustrate performance gains of our method to the classic power method and the truncated power method devised for sparse principal component analysis.

Yuxuan Ye, Jun Han, Ao Hu et al.

End-to-end LLM trading agents have moved quickly from research curiosity to a small ecosystem of named systems, including FinCon, FinMem, TradingAgents, FinAgent, QuantAgent, and FLAG-Trader. Several of these report headline Sharpe ratios that would be material if read at face value on a deployment desk, and associated benchmarks such as FinBen report trading-task Sharpe statistics in the same range. The gap between architecture research and deployment claim has been crossed too freely on both sides of the academia--industry divide. We take a position on that gap: reported alpha from end-to-end LLM trading agents should not be treated as deployment evidence. Before such returns can support claims of deployable trading capability, they must survive structural validity tests for temporal integrity, real-world frictions, counterfactual robustness, predictive calibration, numerical execution, and multi-agent disaggregation. Current public evidence cannot yet distinguish robust predictive ability from temporal contamination, unmodeled frictions, short-window Sharpe uncertainty, narrative fitting, and parametric priors. The problem is not only evaluative but structural. Language confidence is not tradable probability, narrative reasoning is not numerical execution, and model priors may become undisclosed implicit factor exposures. We contribute a minimum reporting protocol suite, P1--P6, with tiered applicability by claim strength, and a conservative modular alternative that uses LLMs as auditable information interfaces upstream of independent calibration, risk, and execution modules. Code and reproduction harness: \url{https://github.com/hj1650782738/Trading}.

Tung Sum Thomas Kwok, Qian Qian, Xiaofeng Lin et al.

Large Language Models exhibit strong reasoning and semantic understanding capabilities but often hallucinate in domains that require expert knowledge, among which fabrications, the generation of factually incorrect yet fluent statements, pose the greatest risk in medical contexts. Existing medical hallucination datasets inadequately capture fabrication phenomena due to limited fabrication coverage, stylistic disparities between human and LLM-authored texts, and distributional drift during hallucinated sample synthesis. To address this, we propose a data-centric pipeline to generate realistic and word-level fabrications that preserve syntactic and stylistic fidelity while introducing subtle factual deviations, resulting in MedFabric. Building upon this dataset, we introduce ETHER, a modular word-level fabrication detector integrating Text2Table Decomposition, Word Masking and Filling and Hybrid Sentence Pair Evaluation to enhance factual alignment. Empirical results demonstrate that MedFabric outperforms state-of-the-art detectors by over 15% on word-level fabrication benchmarks while maintaining consistent performance across structural similarities, offering a comprehensive framework for reliable and domain-specific factuality detection.

Zhaoqiang Liu, Jun Han

In this paper, we propose projected gradient descent (PGD) algorithms for signal estimation from noisy nonlinear measurements. We assume that the unknown $p$-dimensional signal lies near the range of an $L$-Lipschitz continuous generative model with bounded $k$-dimensional inputs. In particular, we consider two cases when the nonlinear link function is either unknown or known. For unknown nonlinearity, similarly to \cite{liu2020generalized}, we make the assumption of sub-Gaussian observations and propose a linear least-squares estimator. We show that when there is no representation error and the sensing vectors are Gaussian, roughly $O(k \log L)$ samples suffice to ensure that a PGD algorithm converges linearly to a point achieving the optimal statistical rate using arbitrary initialization. For known nonlinearity, we assume monotonicity as in \cite{yang2016sparse}, and make much weaker assumptions on the sensing vectors and allow for representation error. We propose a nonlinear least-squares estimator that is guaranteed to enjoy an optimal statistical rate. A corresponding PGD algorithm is provided and is shown to also converge linearly to the estimator using arbitrary initialization. In addition, we present experimental results on image datasets to demonstrate the performance of our PGD algorithms.

Jun Han, Hui Zhang, Zhengru Zhang

In this paper, we focus on the numerical simulation of phase separation about macromolecule microsphere composite (MMC) hydrogel. The model equation is based on Time-Dependent Ginzburg-Landau (TDGL) equation with reticular free energy. We have put forward two $L^2$ stable schemes to simulate simplified TDGL equation. In numerical experiments, we observe that simulating the whole process of phase separation requires a considerably long time. We also notice that the total free energy changes significantly in initial time and varies slightly in the following time. Based on these properties, we introduce an adaptive strategy based on one of stable scheme mentioned. It is found that the introduction of the time adaptivity cannot only resolve the dynamical changes of the solution accurately but also can significantly save CPU time for the long time simulation.

Zhi Yang, Lingfeng Zeng, Fangqi Lou et al.

Multimodal large language models are playing an increasingly significant role in empowering the financial domain, however, the challenges they face, such as multimodal and high-density information and cross-modal multi-hop reasoning, go beyond the evaluation scope of existing multimodal benchmarks. To address this gap, we propose UniFinEval, the first unified multimodal benchmark designed for high-information-density financial environments, covering text, images, and videos. UniFinEval systematically constructs five core financial scenarios grounded in real-world financial systems: Financial Statement Auditing, Company Fundamental Reasoning, Industry Trend Insights, Financial Risk Sensing, and Asset Allocation Analysis. We manually construct a high-quality dataset consisting of 3,767 question-answer pairs in both chinese and english and systematically evaluate 10 mainstream MLLMs under Zero-Shot and CoT settings. Results show that Gemini-3-pro-preview achieves the best overall performance, yet still exhibits a substantial gap compared to financial experts. Further error analysis reveals systematic deficiencies in current models. UniFinEval aims to provide a systematic assessment of MLLMs' capabilities in fine-grained, high-information-density financial environments, thereby enhancing the robustness of MLLMs applications in real-world financial scenarios. Data and code are available at https://github.com/aifinlab/UniFinEval.

Dongxu Zhang, Zhichao Yang, Sepehr Janghorbani et al.

On-policy distillation (OPD), which samples trajectories from the student model and supervises them with a teacher at the token level, avoids relying solely on verifiable terminal rewards and can yield better generalization than off-policy distillation. However, OPD requires expensive on-the-fly sampling of the student policy during training, which substantially increases training cost, especially for long responses. Our initial analysis shows that, during OPD, training signals are often concentrated in the prefix of each output, and that even a short teacher-generated prefix can significantly help the student produce the correct answer. Motivated by these observations, we propose a simple yet effective modification of OPD: we apply the distillation objective only to prefixes of student-generated outputs and terminate each sampling early during distillation. Experiments on a suite of AI-for-Math and out-of-domain benchmarks show that on-policy prefix distillation matches the performance of full OPD while reducing training FLOP by 2x-47x.

Huacan Wang, Chaofa Yuan, Xialie Zhuang et al.

General-purpose technologies reshape economies less by improving individual tools than by enabling new ways to organize production and coordination. We believe AI agents are approaching a similar inflection point: as foundation models make broad task execution and tool use increasingly accessible, the binding constraint shifts from raw capability to how work is delegated, verified, and rewarded at scale. We introduce EpochX, a credits-native marketplace infrastructure for human-agent production networks. EpochX treats humans and agents as peer participants who can post tasks or claim them. Claimed tasks can be decomposed into subtasks and executed through an explicit delivery workflow with verification and acceptance. Crucially, EpochX is designed so that each completed transaction can produce reusable ecosystem assets, including skills, workflows, execution traces, and distilled experience. These assets are stored with explicit dependency structure, enabling retrieval, composition, and cumulative improvement over time. EpochX also introduces a native credit mechanism to make participation economically viable under real compute costs. Credits lock task bounties, budget delegation, settle rewards upon acceptance, and compensate creators when verified assets are reused. By formalizing the end-to-end transaction model together with its asset and incentive layers, EpochX reframes agentic AI as an organizational design problem: building infrastructures where verifiable work leaves persistent, reusable artifacts, and where value flows support durable human-agent collaboration.

Jun Han, Josue Nassar, Sanjit Singh Batra et al.

Machine learning holds great promise for advancing the field of medicine, with electronic health records (EHRs) serving as a primary data source. However, EHRs are often sparse and contain missing data due to various challenges and limitations in data collection and sharing between healthcare providers. Existing techniques for imputing missing values predominantly focus on known unknowns, such as missing or unavailable values of lab test results; most do not explicitly address situations where it is difficult to distinguish what is missing. For instance, a missing diagnosis code in an EHR could signify either that the patient has not been diagnosed with the condition or that a diagnosis was made, but not shared by a provider. Such situations fall into the paradigm of unknown unknowns. To address this challenge, we develop a general purpose algorithm for denoising data to recover unknown missing values in binary EHRs. We design a transformer-based denoising neural network where the output is thresholded adaptively to recover values in cases where we predict data are missing. Our results demonstrate improved accuracy in denoising medical codes within a real EHR dataset compared to existing imputation approaches and leads to increased performance on downstream tasks using the denoised data. In particular, when applying our method to a real world application, predicting hospital readmission from EHRs, our method achieves statistically significant improvement over all existing baselines.

Zhichao Yang, Sepehr Janghorbani, Dongxu Zhang et al.

Rubrics are essential for evaluating open-ended LLM responses, especially in safety-critical domains such as healthcare. However, creating high-quality and domain-specific rubrics typically requires significant human expertise time and development cost, making rubric-based evaluation and training difficult to scale. In this work, we introduce Health-SCORE, a generalizable and scalable rubric-based training and evaluation framework that substantially reduces rubric development costs without sacrificing performance. We show that Health-SCORE provides two practical benefits beyond standalone evaluation: it can be used as a structured reward signal to guide reinforcement learning with safety-aware supervision, and it can be incorporated directly into prompts to improve response quality through in-context learning. Across open-ended healthcare tasks, Health-SCORE achieves evaluation quality comparable to human-created rubrics while significantly lowering development effort, making rubric-based evaluation and training more scalable.

Xuefei Wang, Jialu Wang, Fengbo Zhang et al.

Multi-agent systems (MAS) powered by large language models (LLMs) have emerged as a powerful paradigm for complex problem solving, where performance critically depends on the underlying inter-agent communication topology. However, existing topology generation methods mainly optimize for isolated tasks, while real-world deployments involve streams of evolving tasks, requiring previously effective collaboration patterns to be retained and reused rather than rediscovered or overwritten. We identify a previously underexplored failure mode, \emph{topology forgetting}, in which adapting to new tasks shifts the topology generator away from communication structures required by earlier tasks. This issue stems from cross-task misalignment in both agent-level functional semantics and relational communication structures. To address this challenge, we propose \textbf{\textsc{MasFACT}}, a geometry-aware posterior transfer framework that preserves and reuses historical collaboration knowledge as transferable topology priors. We transfer these priors across task-specific agent spaces through Fused Gromov-Wasserstein optimal transport and perform PAC-Bayes-guided conservative posterior adaptation to balance task-specific plasticity with structural stability. Experiments across class-, domain-, and task-level continual settings demonstrate that \textsc{MasFACT} consistently improves average accuracy while reducing topology forgetting compared to strong topology generation and replay-based baselines, and can be seamlessly integrated with different MAS topology generators.

Jun Han, Shuo Zhang, Wei Li et al.

Financial markets are noisy and non-stationary, making alpha mining highly sensitive to noise in backtesting results and sudden market regime shifts. While recent agentic frameworks improve alpha mining automation, they often lack controllable multi-round search and reliable reuse of validated experience. To address these challenges, we propose QuantaAlpha, an evolutionary alpha mining framework that treats each end-to-end mining run as a trajectory and improves factors through trajectory-level mutation and crossover operations. QuantaAlpha localizes suboptimal steps in each trajectory for targeted revision and recombines complementary high-reward segments to reuse effective patterns, enabling structured exploration and refinement across mining iterations. During factor generation, QuantaAlpha enforces semantic consistency across the hypothesis, factor expression, and executable code, while constraining the complexity and redundancy of the generated factor to mitigate crowding. Extensive experiments on the China Securities Index 300 (CSI 300) demonstrate consistent gains over strong baseline models and prior agentic systems. When utilizing GPT-5.2, QuantaAlpha achieves an Information Coefficient (IC) of 0.1501, with an Annualized Rate of Return (ARR) of 27.75% and a Maximum Drawdown (MDD) of 7.98%. Moreover, factors mined on CSI 300 transfer effectively to the China Securities Index 500 (CSI 500) and the Standard & Poor's 500 Index (S&P 500), delivering 160% and 137% cumulative excess return over four years, respectively, which indicates strong robustness of QuantaAlpha under market distribution shifts.

Chongke Bi, Xin Gao, Jiangkang Deng et al.

Large-scale scientific simulations require significant resources to generate high-resolution time-varying data (TVD). While super-resolution is an efficient post-processing strategy to reduce costs, existing methods rely on a large amount of HR training data, limiting their applicability to diverse simulation scenarios. To address this constraint, we proposed CD-TVD, a novel framework that combines contrastive learning and an improved diffusion-based super-resolution model to achieve accurate 3D super-resolution from limited time-step high-resolution data. During pre-training on historical simulation data, the contrastive encoder and diffusion superresolution modules learn degradation patterns and detailed features of high-resolution and low-resolution samples. In the training phase, the improved diffusion model with a local attention mechanism is fine-tuned using only one newly generated high-resolution timestep, leveraging the degradation knowledge learned by the encoder. This design minimizes the reliance on large-scale high-resolution datasets while maintaining the capability to recover fine-grained details. Experimental results on fluid and atmospheric simulation datasets confirm that CD-TVD delivers accurate and resource-efficient 3D super-resolution, marking a significant advancement in data augmentation for large-scale scientific simulations. The code is available at https://github.com/Xin-Gao-private/CD-TVD.

Lingfeng Zeng, Fangqi Lou, Zixuan Wang et al.

The booming development of AI agents presents unprecedented opportunities for automating complex tasks across various domains. However, their multi-step, multi-tool collaboration capabilities in the financial sector remain underexplored. This paper introduces FinGAIA, an end-to-end benchmark designed to evaluate the practical abilities of AI agents in the financial domain. FinGAIA comprises 407 meticulously crafted tasks, spanning seven major financial sub-domains: securities, funds, banking, insurance, futures, trusts, and asset management. These tasks are organized into three hierarchical levels of scenario depth: basic business analysis, asset decision support, and strategic risk management. We evaluated 10 mainstream AI agents in a zero-shot setting. The best-performing agent, ChatGPT, achieved an overall accuracy of 48.9\%, which, while superior to non-professionals, still lags financial experts by over 35 percentage points. Error analysis has revealed five recurring failure patterns: Cross-modal Alignment Deficiency, Financial Terminological Bias, Operational Process Awareness Barrier, among others. These patterns point to crucial directions for future research. Our work provides the first agent benchmark closely related to the financial domain, aiming to objectively assess and promote the development of agents in this crucial field. Partial data is available at https://github.com/SUFE-AIFLM-Lab/FinGAIA.

Yuhang Zang, Wei Li, Jun Han et al.

Recent Multimodal Large Language Models (MLLMs) are remarkable in vision-language tasks, such as image captioning and question answering, but lack the essential perception ability, i.e., object detection. In this work, we address this limitation by introducing a novel research problem of contextual object detection -- understanding visible objects within different human-AI interactive contexts. Three representative scenarios are investigated, including the language cloze test, visual captioning, and question answering. Moreover, we present ContextDET, a unified multimodal model that is capable of end-to-end differentiable modeling of visual-language contexts, so as to locate, identify, and associate visual objects with language inputs for human-AI interaction. Our ContextDET involves three key submodels: (i) a visual encoder for extracting visual representations, (ii) a pre-trained LLM for multimodal context decoding, and (iii) a visual decoder for predicting bounding boxes given contextual object words. The new generate-then-detect framework enables us to detect object words within human vocabulary. Extensive experiments show the advantages of ContextDET on our proposed CODE benchmark, open-vocabulary detection, and referring image segmentation. Github: https://github.com/yuhangzang/ContextDET.

Jun Han, Zixiang Chen, Yongqian Li et al.

Electronic health records (EHRs) are a pivotal data source that enables numerous applications in computational medicine, e.g., disease progression prediction, clinical trial design, and health economics and outcomes research. Despite wide usability, their sensitive nature raises privacy and confidentially concerns, which limit potential use cases. To tackle these challenges, we explore the use of generative models to synthesize artificial, yet realistic EHRs. While diffusion-based methods have recently demonstrated state-of-the-art performance in generating other data modalities and overcome the training instability and mode collapse issues that plague previous GAN-based approaches, their applications in EHR generation remain underexplored. The discrete nature of tabular medical code data in EHRs poses challenges for high-quality data generation, especially for continuous diffusion models. To this end, we introduce a novel tabular EHR generation method, EHR-D3PM, which enables both unconditional and conditional generation using the discrete diffusion model. Our experiments demonstrate that EHR-D3PM significantly outperforms existing generative baselines on comprehensive fidelity and utility metrics while maintaining less attribute and membership vulnerability risks. Furthermore, we show EHR-D3PM is effective as a data augmentation method and enhances performance on downstream tasks when combined with real data.

Kaiyuan Tang, Kuangshi Ai, Jun Han et al.

Advancements in volume visualization (VolVis) focus on extracting insights from 3D volumetric data by generating visually compelling renderings that reveal complex internal structures. Existing VolVis approaches have explored non-photorealistic rendering techniques to enhance the clarity, expressiveness, and informativeness of visual communication. While effective, these methods often rely on complex predefined rules and are limited to transferring a single style, restricting their flexibility. To overcome these limitations, we advocate the representation of VolVis scenes using differentiable Gaussian primitives combined with pretrained large models to enable arbitrary style transfer and real-time rendering. However, conventional 3D Gaussian primitives tightly couple geometry and appearance, leading to suboptimal stylization results. To address this, we introduce TexGS-VolVis, a textured Gaussian splatting framework for VolVis. TexGS-VolVis employs 2D Gaussian primitives, extending each Gaussian with additional texture and shading attributes, resulting in higher-quality, geometry-consistent stylization and enhanced lighting control during inference. Despite these improvements, achieving flexible and controllable scene editing remains challenging. To further enhance stylization, we develop image- and text-driven non-photorealistic scene editing tailored for TexGS-VolVis and 2D-lift-3D segmentation to enable partial editing with fine-grained control. We evaluate TexGS-VolVis both qualitatively and quantitatively across various volume rendering scenes, demonstrating its superiority over existing methods in terms of efficiency, visual quality, and editing flexibility.

Kichang Lee, Yujin Shin, Jonghyuk Yun et al.

Federated Learning (FL) enables collaborative model training across distributed devices while preserving local data privacy, making it ideal for mobile and embedded systems. However, the decentralized nature of FL also opens vulnerabilities to model poisoning attacks, particularly backdoor attacks, where adversaries implant trigger patterns to manipulate model predictions. In this paper, we propose DeTrigger, a scalable and efficient backdoor-robust federated learning framework that leverages insights from adversarial attack methodologies. By employing gradient analysis with temperature scaling, DeTrigger detects and isolates backdoor triggers, allowing for precise model weight pruning of backdoor activations without sacrificing benign model knowledge. Extensive evaluations across four widely used datasets demonstrate that DeTrigger achieves up to 251x faster detection than traditional methods and mitigates backdoor attacks by up to 98.9%, with minimal impact on global model accuracy. Our findings establish DeTrigger as a robust and scalable solution to protect federated learning environments against sophisticated backdoor threats.

Connor Bolton, Yan Long, Jun Han et al.

Smartphone motion sensors provide a concealed mechanism for eavesdropping on acoustic information, like touchtones, emitted by a device. Eavesdropping on touchtones exposes credit card information, banking pins, and social security card numbers to malicious 3rd party apps requiring only motion sensor data. This paper's primary contribution is an analysis rooted in physics and signal processing theory of several eavesdropping mitigations, which could be implemented in a smartphone update. We verify our analysis imperially to show how previously suggested mitigations, i.e. a low-pass filter, can undesirably reduce the motion sensor data to all applications by 83% but only reduce an advanced adversary's accuracy by less than one percent. Other designs, i.e. anti-aliasing filters, can fully preserve the motion sensor data to support benign application functionality while reducing attack accuracy by 50.1%. We intend for this analysis to motivate the need for deployable mitigations against acoustic leakage on smartphone motion sensors, including but not limited to touchtones, while also providing a basis for future mitigations to improve upon.

Zhaoqiang Liu, Subhroshekhar Ghosh, Jun Han et al.

In 1-bit compressive sensing, each measurement is quantized to a single bit, namely the sign of a linear function of an unknown vector, and the goal is to accurately recover the vector. While it is most popular to assume a standard Gaussian sensing matrix for 1-bit compressive sensing, using structured sensing matrices such as partial Gaussian circulant matrices is of significant practical importance due to their faster matrix operations. In this paper, we provide recovery guarantees for a correlation-based optimization algorithm for robust 1-bit compressive sensing with randomly signed partial Gaussian circulant matrices and generative models. Under suitable assumptions, we match guarantees that were previously only known to hold for i.i.d.~Gaussian matrices that require significantly more computation. We make use of a practical iterative algorithm, and perform numerical experiments on image datasets to corroborate our theoretical results.

Kai Feng, Weixing Li, Jun Han et al.

Rotating object detection has wide applications in aerial photographs, remote sensing images, UAVs, etc. At present, most of the rotating object detection datasets focus on the field of remote sensing, and these images are usually shot in high-altitude scenes. However, image datasets captured at low-altitude areas also should be concerned, such as drone-based datasets. So we present a low-altitude dronebased dataset, named UAV-ROD, aiming to promote the research and development in rotating object detection and UAV applications. The UAV-ROD consists of 1577 images and 30,090 instances of car category annotated by oriented bounding boxes. In particular, The UAV-ROD can be utilized for the rotating object detection, vehicle orientation recognition and object counting tasks. Compared with horizontal object detection, the regression stage of the rotation detection is a tricky problem. In this paper, we propose a rotating object detector TS4Net, which contains anchor refinement module (ARM) and two-stage sample selective strategy (TS4). The ARM can convert preseted horizontal anchors into high-quality rotated anchors through twostage anchor refinement. The TS4 module utilizes different constrained sample selective strategies to allocate positive and negative samples, which is adaptive to the regression task in different stages. Benefiting from the ARM and TS4, the TS4Net can achieve superior performance for rotating object detection solely with one preseted horizontal anchor. Extensive experimental results on UAV-ROD dataset and three remote sensing datasets DOTA, HRSC2016 and UCAS-AOD demonstrate that our method achieves competitive performance against most state-of-the-art methods.

Hao Zheng, Jun Han, Hongxiao Wang et al.

A large labeled dataset is a key to the success of supervised deep learning, but for medical image segmentation, it is highly challenging to obtain sufficient annotated images for model training. In many scenarios, unannotated images are abundant and easy to acquire. Self-supervised learning (SSL) has shown great potentials in exploiting raw data information and representation learning. In this paper, we propose Hierarchical Self-Supervised Learning (HSSL), a new self-supervised framework that boosts medical image segmentation by making good use of unannotated data. Unlike the current literature on task-specific self-supervised pretraining followed by supervised fine-tuning, we utilize SSL to learn task-agnostic knowledge from heterogeneous data for various medical image segmentation tasks. Specifically, we first aggregate a dataset from several medical challenges, then pre-train the network in a self-supervised manner, and finally fine-tune on labeled data. We develop a new loss function by combining contrastive loss and classification loss and pretrain an encoder-decoder architecture for segmentation tasks. Our extensive experiments show that multi-domain joint pre-training benefits downstream segmentation tasks and outperforms single-domain pre-training significantly. Compared to learning from scratch, our new method yields better performance on various tasks (e.g., +0.69% to +18.60% in Dice scores with 5% of annotated data). With limited amounts of training data, our method can substantially bridge the performance gap w.r.t. denser annotations (e.g., 10% vs.~100% of annotated data).

Mikhail Fomichev, Julia Hesse, Lars Almon et al.

With the advent of the Internet of Things (IoT), establishing a secure channel between smart devices becomes crucial. Recent research proposes zero-interaction pairing (ZIP), which enables pairing without user assistance by utilizing devices' physical context (e.g., ambient audio) to obtain a shared secret key. The state-of-the-art ZIP schemes suffer from three limitations: (1) prolonged pairing time (i.e., minutes or hours), (2) vulnerability to brute-force offline attacks on a shared key, and (3) susceptibility to attacks caused by predictable context (e.g., replay attack) because they rely on limited entropy of physical context to protect a shared key. We address these limitations, proposing FastZIP, a novel ZIP scheme that significantly reduces pairing time while preventing offline and predictable context attacks. In particular, we adapt a recently introduced Fuzzy Password-Authenticated Key Exchange (fPAKE) protocol and utilize sensor fusion, maximizing their advantages. We instantiate FastZIP for intra-car device pairing to demonstrate its feasibility and show how the design of FastZIP can be adapted to other ZIP use cases. We implement FastZIP and evaluate it by driving four cars for a total of 800 km. We achieve up to three times shorter pairing time compared to the state-of-the-art ZIP schemes while assuring robust security with adversarial error rates below 0.5%.

Jun Han, Martin Renqiang Min, Ligong Han et al.

Learning disentangled representations leads to interpretable models and facilitates data generation with style transfer, which has been extensively studied on static data such as images in an unsupervised learning framework. However, only a few works have explored unsupervised disentangled sequential representation learning due to challenges of generating sequential data. In this paper, we propose recurrent Wasserstein Autoencoder (R-WAE), a new framework for generative modeling of sequential data. R-WAE disentangles the representation of an input sequence into static and dynamic factors (i.e., time-invariant and time-varying parts). Our theoretical analysis shows that, R-WAE minimizes an upper bound of a penalized form of the Wasserstein distance between model distribution and sequential data distribution, and simultaneously maximizes the mutual information between input data and different disentangled latent factors, respectively. This is superior to (recurrent) VAE which does not explicitly enforce mutual information maximization between input data and disentangled latent representations. When the number of actions in sequential data is available as weak supervision information, R-WAE is extended to learn a categorical latent representation of actions to improve its disentanglement. Experiments on a variety of datasets show that our models outperform other baselines with the same settings in terms of disentanglement and unconditional video generation both quantitatively and qualitatively.

Muhammad Ashad Kabir, Jun Han, Md. Arafat Hossain et al.

Software behavioral models have proven useful for emulating and testing software systems. Many techniques have been proposed to infer behavioral models of software systems from their interaction traces. The quality of the inferred model is critical to its successful use. While generalization is necessary to deduce concise behavioral models, existing techniques of inferring models, in general, overgeneralize what behavior is valid. Imprecise models include many spurious behaviors, and thus compromise the effectiveness of their use. In this paper, we propose a novel technique that increases the accuracy of the behavioral model inferred from interaction traces. The essence of our approach is a heuristic-based generalization and truthful minimization. The set of heuristics include patterns to match input traces and generalize them towards concise model representations. Furthermore, we adopt a truthful minimization technique to merge these generalized traces. The key insight of our approach is to infer a concise behavioral model without compromising its accuracy. We present an empirical evaluation of how our approach improves upon the state-of-the-art specification inference techniques. The results show that our approach mines model with 100% precision and recall with a limited computation overhead.

Madhumitha Harishankar, Jun Han, Sai Vineeth Kalluru Srinivas et al.

The resolution of GPS measurements, especially in urban areas, is insufficient for identifying a vehicle's lane. In this work, we develop a deep LSTM neural network model LaNet that determines the lane vehicles are on by periodically classifying accelerometer samples collected by vehicles as they drive in real time. Our key finding is that even adjacent patches of road surfaces contain characteristics that are sufficiently unique to differentiate between lanes, i.e., roads inherently exhibit differing bumps, cracks, potholes, and surface unevenness. Cars can capture this road surface information as they drive using inexpensive, easy-to-install accelerometers that increasingly come fitted in cars and can be accessed via the CAN-bus. We collect an aggregate of 60 km driving data and synthesize more based on this that capture factors such as variable driving speed, vehicle suspensions, and accelerometer noise. Our formulated LSTM-based deep learning model, LaNet, learns lane-specific sequences of road surface events (bumps, cracks etc.) and yields 100% lane classification accuracy with 200 meters of driving data, achieving over 90% with just 100 m (correspondingly to roughly one minute of driving). We design the LaNet model to be practical for use in real-time lane classification and show with extensive experiments that LaNet yields high classification accuracy even on smooth roads, on large multi-lane roads, and on drives with frequent lane changes. Since different road surfaces have different inherent characteristics or entropy, we excavate our neural network model and discover a mechanism to easily characterize the achievable classification accuracies in a road over various driving distances by training the model just once. We present LaNet as a low-cost, easily deployable and highly accurate way to achieve fine-grained lane identification.

Jun Han

Approximate inference in probability models is a fundamental task in machine learning. Approximate inference provides powerful tools to Bayesian reasoning, decision making, and Bayesian deep learning. The main goal is to estimate the expectation of interested functions w.r.t. a target distribution. When it comes to high dimensional probability models and large datasets, efficient approximate inference becomes critically important. In this thesis, we propose a new framework for approximate inference, which combines the advantages of these three frameworks and overcomes their limitations. Our proposed four algorithms are motivated by the recent computational progress of Stein's method. Our proposed algorithms are applied to continuous and discrete distributions under the setting when the gradient information of the target distribution is available or unavailable. Theoretical analysis is provided to prove the convergence of our proposed algorithms. Our adaptive IS algorithm iteratively improves the importance proposal by functionally decreasing the KL divergence between the updated proposal and the target. When the gradient of the target is unavailable, our proposed sampling algorithm leverages the gradient of a surrogate model and corrects induced bias with importance weights, which significantly outperforms other gradient-free sampling algorithms. In addition, our theoretical results enable us to perform the goodness-of-fit test on discrete distributions. At the end of the thesis, we propose an importance-weighted method to efficiently aggregate local models in distributed learning with one-shot communication. Results on simulated and real datasets indicate the statistical efficiency and wide applicability of our algorithm.

Jun Han, Fan Ding, Xianglong Liu et al.

Gradient-based approximate inference methods, such as Stein variational gradient descent (SVGD), provide simple and general-purpose inference engines for differentiable continuous distributions. However, existing forms of SVGD cannot be directly applied to discrete distributions. In this work, we fill this gap by proposing a simple yet general framework that transforms discrete distributions to equivalent piecewise continuous distributions, on which the gradient-free SVGD is applied to perform efficient approximate inference. The empirical results show that our method outperforms traditional algorithms such as Gibbs sampling and discontinuous Hamiltonian Monte Carlo on various challenging benchmarks of discrete graphical models. We demonstrate that our method provides a promising tool for learning ensembles of binarized neural network (BNN), outperforming other widely used ensemble methods on learning binarized AlexNet on CIFAR-10 dataset. In addition, such transform can be straightforwardly employed in gradient-free kernelized Stein discrepancy to perform goodness-of-fit (GOF) test on discrete distributions. Our proposed method outperforms existing GOF test methods for intractable discrete distributions.

Indika Kumara, Jun Han, Alan Colman et al.

In a multi-tenant service network, multiple virtual service networks (VSNs), one for each tenant, coexist on the same service network. The tenants themselves need to be able to dynamically create and customize their own VSNs to support their initial and changing functional and performance requirements. These tasks are problematic for them due to: 1) platform-specific knowledge required, 2) the existence of a large number of customization options and their dependencies, and 3) the complexity in deriving the right subset of options. In this paper, we present an approach to enable and simplify the tenant-driven customization of multi-tenant service networks. We propose to use feature as a high-level customization abstraction. A regulated collaboration among a set of services in the service network realizes a feature. A software engineer can design a customization policy for a service network using the mappings between features and collaborations, and enact the policy with the controller of the service network. A tenant can then specify the requirements for its VSN as a set of functional and performance features. A customization request from a tenant triggers the customization policy of the service network, which (re)configures the corresponding VSN at runtime to realize the selected features. We show the feasibility of our approach with two case studies and a performance evaluation.

Iqbal H. Sarker, Alan Colman, Jun Han et al.

This paper formulates the problem of building a context-aware predictive model based on user diverse behavioral activities with smartphones. In the area of machine learning and data science, a tree-like model as that of decision tree is considered as one of the most popular classification techniques, which can be used to build a data-driven predictive model. The traditional decision tree model typically creates a number of leaf nodes as decision nodes that represent context-specific rigid decisions, and consequently may cause overfitting problem in behavior modeling. However, in many practical scenarios within the context-aware environment, the generalized outcomes could play an important role to effectively capture user behavior. In this paper, we propose a behavioral decision tree, "BehavDT" context-aware model that takes into account user behavior-oriented generalization according to individual preference level. The BehavDT model outputs not only the generalized decisions but also the context-specific decisions in relevant exceptional cases. The effectiveness of our BehavDT model is studied by conducting experiments on individual user real smartphone datasets. Our experimental results show that the proposed BehavDT context-aware model is more effective when compared with the traditional machine learning approaches, in predicting user diverse behaviors considering multi-dimensional contexts.

Iqbal H. Sarker, Alan Colman, Jun Han et al.

The electronic calendar is a valuable resource nowadays for managing our daily life appointments or schedules, also known as events, ranging from professional to highly personal. Researchers have studied various types of calendar events to predict smartphone user behavior for incoming mobile communications. However, these studies typically do not take into account behavioral variations between individuals. In the real world, smartphone users can differ widely from each other in how they respond to incoming communications during their scheduled events. Moreover, an individual user may respond the incoming communications differently in different contexts subject to what type of event is scheduled in her personal calendar. Thus, a static calendar-based behavioral model for individual smartphone users does not necessarily reflect their behavior to the incoming communications. In this paper, we present a machine learning based context-aware model that is personalized and dynamically identifies individual's dominant behavior for their scheduled events using logged time-series smartphone data, and shortly name as ``CalBehav''. The experimental results based on real datasets from calendar and phone logs, show that this data-driven personalized model is more effective for intelligently managing the incoming mobile communications compared to existing calendar-based approaches.

Chao Qin, Haoyang Ye, Christian E. Pranata et al.

We present LINS, a lightweight lidar-inertial state estimator, for real-time ego-motion estimation. The proposed method enables robust and efficient navigation for ground vehicles in challenging environments, such as feature-less scenes, via fusing a 6-axis IMU and a 3D lidar in a tightly-coupled scheme. An iterated error-state Kalman filter (ESKF) is designed to correct the estimated state recursively by generating new feature correspondences in each iteration, and to keep the system computationally tractable. Moreover, we use a robocentric formulation that represents the state in a moving local frame in order to prevent filter divergence in a long run. To validate robustness and generalizability, extensive experiments are performed in various scenarios. Experimental results indicate that LINS offers comparable performance with the state-of-the-art lidar-inertial odometry in terms of stability and accuracy and has order-of-magnitude improvement in speed.

Nitya Lakshmanan, Inkyu Bang, Min Suk Kang et al.

The proliferation of surveillance cameras has greatly improved the physical security of many security-critical properties including buildings, stores, and homes. However, recent surveillance camera looping attacks demonstrate new security threats - adversaries can replay a seemingly benign video feed of a place of interest while trespassing or stealing valuables without getting caught. Unfortunately, such attacks are extremely difficult to detect in real-time due to cost and implementation constraints. In this paper, we propose SurFi to detect these attacks in real-time by utilizing commonly available Wi-Fi signals. In particular, we leverage that channel state information (CSI) from Wi-Fi signals also perceives human activities in the place of interest in addition to surveillance cameras. SurFi processes and correlates the live video feeds and the Wi-Fi CSI signals to detect any mismatches that would identify the presence of the surveillance camera looping attacks. SurFi does not require the deployment of additional infrastructure because Wi-Fi transceivers are easily found in the urban indoor environment. We design and implement the SurFi system and evaluate its effectiveness in detecting surveillance camera looping attacks. Our evaluation demonstrates that SurFi effectively identifies attacks with up to an attack detection accuracy of 98.8% and 0.1% false positive rate

Quan Z. Sheng, Jian Yu, Hanchuan Xu et al.

In the era of Web of Things and Services, Context-aware Web Services (CASs) are emerging as an important technology for building innovative context-aware applications. CASs enable the information integration from both the physical and virtual world, which affects human living. However, it is challenging to build CASs, due to the lack of context provisioning management approach and limited generic approach for formalizing the development process. We therefore propose ContextServ, a platform that uses a model-driven approach to support the full life cycle of CASs development, hence offering significant design and management flexibility. ContextServ implements a proposed UML-based modelling language ContextUML to support multiple modelling languages. It also supports dynamic adaptation of WS-BPEL based context-aware composite services by weaving context-aware rules into the process. Extensive experimental evaluations on ContextServ and its components showcase that ContextServ can support effective development and efficient execution of context-aware Web services.

Iqbal H. Sarker, Alan Colman, MA Kabir et al.

Mobile phones can record individual's daily behavioral data as a time-series. In this paper, we present an effective time-series segmentation technique that extracts optimal time segments of individual's similar behavioral characteristics utilizing their mobile phone data. One of the determinants of an individual's behavior is the various activities undertaken at various times-of-the-day and days-of-the-week. In many cases, such behavior will follow temporal patterns. Currently, researchers use either equal or unequal interval-based segmentation of time for mining mobile phone users' behavior. Most of them take into account static temporal coverage of 24-h-a-day and few of them take into account the number of incidences in time-series data. However, such segmentations do not necessarily map to the patterns of individual user activity and subsequent behavior because of not taking into account the diverse behaviors of individuals over time-of-the-week. Therefore, we propose a behavior-oriented time segmentation (BOTS) technique that takes into account not only the temporal coverage of the week but also the number of incidences of diverse behaviors dynamically for producing similar behavioral time segments over the week utilizing time-series data. Experiments on the real mobile phone datasets show that our proposed segmentation technique better captures the user's dominant behavior at various times-of-the-day and days-of-the-week enabling the generation of high confidence temporal rules in order to mine individual mobile phone users' behavior.

Jun Han, Salvator Lombardo, Christopher Schroers et al.

The usage of deep generative models for image compression has led to impressive performance gains over classical codecs while neural video compression is still in its infancy. Here, we propose an end-to-end, deep generative modeling approach to compress temporal sequences with a focus on video. Our approach builds upon variational autoencoder (VAE) models for sequential data and combines them with recent work on neural image compression. The approach jointly learns to transform the original sequence into a lower-dimensional representation as well as to discretize and entropy code this representation according to predictions of the sequential VAE. Rate-distortion evaluations on small videos from public data sets with varying complexity and diversity show that our model yields competitive results when trained on generic video content. Extreme compression performance is achieved when training the model on specialized content.

Jun Han, Qiang Liu

Stein variational gradient decent (SVGD) has been shown to be a powerful approximate inference algorithm for complex distributions. However, the standard SVGD requires calculating the gradient of the target density and cannot be applied when the gradient is unavailable. In this work, we develop a gradient-free variant of SVGD (GF-SVGD), which replaces the true gradient with a surrogate gradient, and corrects the induced bias by re-weighting the gradients in a proper form. We show that our GF-SVGD can be viewed as the standard SVGD with a special choice of kernel, and hence directly inherits the theoretical properties of SVGD. We shed insights on the empirical choice of the surrogate gradient and propose an annealed GF-SVGD that leverages the idea of simulated annealing to improve the performance on high dimensional complex distributions. Empirical studies show that our method consistently outperforms a number of recent advanced gradient-free MCMC methods.

Iqbal H. Sarker, Muhammad Ashad Kabir, Alan Colman et al.

The presence of noisy instances in mobile phone data is a fundamental issue for classifying user phone call behavior (i.e., accept, reject, missed and outgoing), with many potential negative consequences. The classification accuracy may decrease and the complexity of the classifiers may increase due to the number of redundant training samples. To detect such noisy instances from a training dataset, researchers use naive Bayes classifier (NBC) as it identifies misclassified instances by taking into account independence assumption and conditional probabilities of the attributes. However, some of these misclassified instances might indicate usages behavioral patterns of individual mobile phone users. Existing naive Bayes classifier based noise detection techniques have not considered this issue and, thus, are lacking in classification accuracy. In this paper, we propose an improved noise detection technique based on naive Bayes classifier for effectively classifying users' phone call behaviors. In order to improve the classification accuracy, we effectively identify noisy instances from the training dataset by analyzing the behavioral patterns of individuals. We dynamically determine a noise threshold according to individual's unique behavioral patterns by using both the naive Bayes classifier and Laplace estimator. We use this noise threshold to identify noisy instances. To measure the effectiveness of our technique in classifying user phone call behavior, we employ the most popular classification algorithm (e.g., decision tree). Experimental results on the real phone call log dataset show that our proposed technique more accurately identifies the noisy instances from the training datasets that leads to better classification accuracy.

Zeinab Farahmandpour, Steve Versteeg, Anand Kameswaran et al.

Service virtualization is an approach that uses virtualized environments to automatically test enterprise services in production-like conditions. Many techniques have been proposed to provide such a realistic environment for enterprise services. The Internet-of-Things (IoT) is an emerging field which connects a diverse set of devices over different transport layers, using a variety of protocols. Provisioning a virtual testbed of IoT devices can accelerate IoT application development by enabling automated testing without requiring a continuous connection to the physical devices. One solution is to expand existing enterprise service virtualization to IoT environments. There are various structural differences between the two environments that should be considered to implement appropriate service virtualization for IoT. This paper examines the structural differences between various IoT protocols and enterprise protocols and identifies key technical challenges that need to be addressed to implement service virtualization in IoT environments.

Jun Han, Qiang Liu

We propose a novel adaptive importance sampling algorithm which incorporates Stein variational gradient decent algorithm (SVGD) with importance sampling (IS). Our algorithm leverages the nonparametric transforms in SVGD to iteratively decrease the KL divergence between our importance proposal and the target distribution. The advantages of this algorithm are twofold: first, our algorithm turns SVGD into a standard IS algorithm, allowing us to use standard diagnostic and analytic tools of IS to evaluate and interpret the results; second, we do not restrict the choice of our importance proposal to predefined distribution families like traditional (adaptive) IS methods. Empirical experiments demonstrate that our algorithm performs well on evaluating partition functions of restricted Boltzmann machines and testing likelihood of variational auto-encoders.

A. S. M. Kayes, Jun Han, Wenny Rahayu et al.

In today's dynamic ICT environments, the ability to control users' access to resources becomes ever important. On the one hand, it should adapt to the users' changing needs; on the other hand, it should not be compromised. Therefore, it is essential to have a flexible access control model, incorporating dynamically changing context information. Towards this end, this paper introduces a policy framework for context-aware access control (CAAC) applications that extends the role-based access control model with both dynamic associations of user-role and role-permission capabilities. We first present a formal model of CAAC policies for our framework. Using this model, we then introduce an ontology-based approach and a software prototype for modelling and enforcing CAAC policies. In addition, we evaluate our policy ontology model and framework by considering (i) the completeness of the ontology concepts, specifying different context-aware user-role and role-permission assignment policies from the healthcare scenarios; (ii) the correctness and consistency of the ontology semantics, assessing the core and domain-specific ontologies through the healthcare case study; and (iii) the performance of the framework by means of response time. The evaluation results demonstrate the feasibility of our framework and quantify the performance overhead of achieving context-aware access control to information resources.

Jun Han, Qiang Liu

In distributed, or privacy-preserving learning, we are often given a set of probabilistic models estimated from different local repositories, and asked to combine them into a single model that gives efficient statistical estimation. A simple method is to linearly average the parameters of the local models, which, however, tends to be degenerate or not applicable on non-convex models, or models with different parameter dimensions. One more practical strategy is to generate bootstrap samples from the local models, and then learn a joint model based on the combined bootstrap set. Unfortunately, the bootstrap procedure introduces additional noise and can significantly deteriorate the performance. In this work, we propose two variance reduction methods to correct the bootstrap noise, including a weighted M-estimator that is both statistically efficient and practically powerful. Both theoretical and empirical analysis is provided to demonstrate our methods.

Cameron Hine, Jean-Guy Schneider, Jun Han et al.

Constructing testbeds for systems which are interconnected with large networks of other software services is a challenging task. It is particularly difficult to create testbeds facilitating evaluation of the non-functional qualities of a system, such as scalability, that can be expected in production deployments. Software service emulation is an approach for creating such testbeds where service behaviour is defined by emulate-able models executed in an emulation runtime environment. We present (i) a meta-modelling framework supporting emulate-able service modelling (including messages, protocol, behaviour and states), and (ii) Kaluta, an emulation environment able to concurrently execute large numbers (thousands) of service models, providing a testbed which mimics the behaviour and characteristics of large networks of interconnected software services. Experiments show that Kaluta can emulate 10,000 servers using a single physical machine, and is a practical testbed for scalability testing of a real, enterprise-grade identity management suite. The insights gained into the tested enterprise system were used to enhance its design.

Steve Versteeg, Miao Du, Jean-Guy Schneider et al.

Large enterprise software systems make many complex interactions with other services in their environment. Developing and testing for production-like conditions is therefore a very challenging task. Current approaches include emulation of dependent services using either explicit modelling or record-and-replay approaches. Models require deep knowledge of the target services while record-and-replay is limited in accuracy. Both face developmental and scaling issues. We present a new technique that improves the accuracy of record-and-replay approaches, without requiring prior knowledge of the service protocols. The approach uses Multiple Sequence Alignment to derive message prototypes from recorded system interactions and a scheme to match incoming request messages against prototypes to generate response messages. We use a modified Needleman-Wunsch algorithm for distance calculation during message matching. Our approach has shown greater than 99% accuracy for four evaluated enterprise system messaging protocols. The approach has been successfully integrated into the CA Service Virtualization commercial product to complement its existing techniques.

Steve Versteeg, Miao Du, John Bird et al.

Service virtualisation is a supporting tool for DevOps to generate interactive service models of dependency systems on which a system-under-test relies. These service models allow applications under development to be continuously tested against production-like conditions. Generating these virtual service models requires expert knowledge of the service protocol, which may not always be available. However, service models may be generated automatically from network traces. Previous work has used the Needleman-Wunsch algorithm to select a response from the service model to play back for a live request. We propose an extension of the Needleman-Wunsch algorithm, which uses entropy analysis to automatically detect the critical matching fields for selecting a response. Empirical tests against four enterprise protocols demonstrate that entropy weighted matching can improve response accuracy.

Miao Du, Steve Versteeg, Jean-Guy Schneider et al.

Enterprise software systems make complex interactions with other services in their environment. Developing and testing for production-like conditions is therefore a challenging task. Prior approaches include emulations of the dependency services using either explicit modelling or record-and-replay approaches. Models require deep knowledge of the target services while record-and-replay is limited in accuracy. We present a new technique that improves the accuracy of record-and-replay approaches, without requiring prior knowledge of the services. The approach uses multiple sequence alignment to derive message prototypes from recorded system interactions and a scheme to match incoming request messages against message prototypes to generate response messages. We introduce a modified Needleman-Wunsch algorithm for distance calculation during message matching, wildcards in message prototypes for high variability sections, and entropy-based weightings in distance calculations for increased accuracy. Combined, our new approach has shown greater than 99% accuracy for four evaluated enterprise system messaging protocols.