Searcharxiv⌕ Search

arXiv subjects

Zimu Zhou

Publications and source records attributed to Zimu Zhou.

At least 19 recordsLinked to original sources
arXiv2026

Practical Evaluation of FFT-Based Thickness Extraction for Thick-Film Reflectometry

Fast Fourier transform (FFT) is widely used for thick-film reflectometry because of its simplicity and computational efficiency. However, its performance under practical thick-film measurement conditions has received limited experimental evaluation. In this work, FFT, Linearized Reflectance Zero-Crossing (LRZ), and optical model fitting were compared using reflectometry measurements from a nominal 52 {\mu}m dielectric film acquired on a production wafer, with White-Light Interferometry (WLI) serving as an independent reference. Although FFT exhibited excellent repeatability, it showed systematic deviation from WLI (RMSE = 0.62 {\mu}m), whereas LRZ significantly improved accuracy (RMSE = 0.20 {\mu}m). Simulations and theoretical analysis indicate that the observed systematic deviation is consistent with spectral leakage caused by finite measurement windows and non-integer fringe periodicity. These results demonstrate that excellent repeatability does not necessarily imply high accuracy and highlight the advantages of zero-crossing-based approaches for thick-film thickness metrology.

physics.optics↗
arXiv2026

MemLens: A Value-Aware Memory Management System with Interactive Analytics for LLM-based Agents

Recently, memory management has become a key infrastructure for LLM-based agents, as it directly affects long-horizon reasoning, personalized responses, and knowledge reuse. However, existing LLM memory systems typically adopt a coarse-grained (utility-agnostic) manner that treats heterogeneous user-LLM interaction records uniformly, leading to redundant and low-impact records persisting in the memory repository. To address this challenge, we present MemLens, a value-aware memory management system that takes memory records as first-class data objects. MemLens provides an end-to-end interactive analytics dashboard that exposes the complete memory lifecycle, including Shapley-style memory evaluation, value-aware storage, and memory-assisted response. Through a study-copilot application, the system enables users to inspect memory values, visualize hierarchical memory structures, and compare various memory management strategies in terms of response quality, retrieval latency, and token consumption. Therefore, our MemLens can serve as an efficient, interpretable, and personalized long-term memory management system for LLM-based agents.

cs.DB↗
arXiv2026

FedMosaic: Federated Retrieval-Augmented Generation via Parametric Adapters

Retrieval-Augmented Generation (RAG) enhances Large Language Models (LLMs) by grounding generation in external knowledge to improve factuality and reduce hallucinations. Yet most deployments assume a centralized corpus, which is infeasible in privacy aware domains where knowledge remains siloed. This motivates federated RAG (FedRAG), where a central LLM server collaborates with distributed silos without sharing raw documents. In context RAG violates this requirement by transmitting verbatim documents, whereas parametric RAG encodes documents into lightweight adapters that merge with a frozen LLM at inference, avoiding raw-text exchange. We adopt the parametric approach but face two unique challenges induced by FedRAG: high storage and communication from per-document adapters, and destructive aggregation caused by indiscriminately merging multiple adapters. We present FedMosaic, the first federated RAG framework built on parametric adapters. FedMosaic clusters semantically related documents into multi-document adapters with document-specific masks to reduce overhead while preserving specificity, and performs selective adapter aggregation to combine only relevance-aligned, nonconflicting adapters. Experiments show that FedMosaic achieves an average 10.9% higher accuracy than state-of-the-art methods in four categories, while lowering storage costs by 78.8% to 86.3% and communication costs by 91.4%, and never sharing raw documents.

cs.CL↗
arXiv2026

Federated Personalization of Early-Exit Networks

Personalized Federated Learning (PFL) excels at tailoring client-specific models, which is particularly critical for decentralized and heterogeneous data environments, yet existing methods produce static models with a fixed tradeoff between accuracy and efficiency. This inherent static nature limits their ability to adapt to inference demands that vary with context and resource availability, posing a challenge for real-world deployment. Early-exit networks (EENs), which enable adaptive inference via intermediate classifiers, offer a promising solution. However, integrating EENs into PFL introduces two intertwined conflicts: client-wise heterogeneity across clients and depth-wise interference arising from conflicting exit objectives. Prior studies fail to resolve both conflicts simultaneously, leading to suboptimal performance. In this paper, we propose X-FED, a novel Conflict-Aware Cross-Client Federated Exit Distillation framework that jointly addresses both client- and depth-wise conflicts while extending PFL to early-exit networks. At its core, X-FED employs a progressive, depth-prioritized student coordination mechanism that mitigates interference among shallow and deep exits while enabling effective personalized knowledge transfer across clients. Furthermore, we introduce a client-decoupled formulation that reduces communication overhead with theoretical soundness. Extensive evaluations on various datasets show that compared to state-of-the-art PFL and PFL-EE methods, X-FED achieves higher accuracy while reducing inference costs by 30.79%-46.86%.

cs.LG↗
arXiv2025

Extending Reflectometry Range, A Zero-Crossing Algorithm for Thick Film Metrology

Accurate and high-efficiency film metrology remains a key challenge in High-Volume Manufacturing (HVM), where conventional spectroscopic reflectometry and white light interferometry (WLI) are either limited by model dependence or throughput. In this work, we extend the measurable film-thickness range of reflectometry to at least 50 um through a new model-free algorithm, the Linearized Reflectance Zero-Crossing (LRZ) method. The approach builds upon the previously reported Linearized Reflectance Extrema (LRE) technique but eliminates the sensitivity to spectral sampling and fringe attenuation that degrade performance in the thick-film regime. By linearizing phase response and extracting zero-crossing positions in wavenumber space, LRZ provides robust and repeatable thickness estimation without iterative fitting, achieving comparable accuracy with much higher computational efficiency than conventional model-based methods. Validation using more than 80 measurements on alumina films over NiFe substrates shows excellent correlation with WLI (r = 0.97) and low gauge repeatability and reproducibility (GR&R < 3%). Moreover, LRZ achieves an average Move-Acquire-Measure (MAM) time of approximately 2 s, which is about 7 times faster than WLI. The proposed method enables fast, accurate, and model-independent optical metrology for thick films, offering a practical solution for advanced HVM process control.

physics.optics↗
arXiv2025

Accurate and Efficient Multivariate Time Series Forecasting via Offline Clustering

Accurate and efficient multivariate time series (MTS) forecasting is essential for applications such as traffic management and weather prediction, which depend on capturing long-range temporal dependencies and interactions between entities. Existing methods, particularly those based on Transformer architectures, compute pairwise dependencies across all time steps, leading to a computational complexity that scales quadratically with the length of the input. To overcome these challenges, we introduce the Forecaster with Offline Clustering Using Segments (FOCUS), a novel approach to MTS forecasting that simplifies long-range dependency modeling through the use of prototypes extracted via offline clustering. These prototypes encapsulate high-level events in the real-world system underlying the data, summarizing the key characteristics of similar time segments. In the online phase, FOCUS dynamically adapts these patterns to the current input and captures dependencies between the input segment and high-level events, enabling both accurate and efficient forecasting. By identifying prototypes during the offline clustering phase, FOCUS reduces the computational complexity of modeling long-range dependencies in the online phase to linear scaling. Extensive experiments across diverse benchmarks demonstrate that FOCUS achieves state-of-the-art accuracy while significantly reducing computational costs.

cs.LG↗
arXiv2025

Efficient Data Valuation Approximation in Federated Learning: A Sampling-based Approach

Federated learning paradigm to utilize datasets across multiple data providers. In FL, cross-silo data providers often hesitate to share their high-quality dataset unless their data value can be fairly assessed. Shapley value (SV) has been advocated as the standard metric for data valuation in FL due to its desirable properties. However, the computational overhead of SV is prohibitive in practice, as it inherently requires training and evaluating an FL model across an exponential number of dataset combinations. Furthermore, existing solutions fail to achieve high accuracy and efficiency, making practical use of SV still out of reach, because they ignore choosing suitable computation scheme for approximation framework and overlook the property of utility function in FL. We first propose a unified stratified-sampling framework for two widely-used schemes. Then, we analyze and choose the more promising scheme under the FL linear regression assumption. After that, we identify a phenomenon termed key combinations, where only limited dataset combinations have a high-impact on final data value. Building on these insights, we propose a practical approximation algorithm, IPSS, which strategically selects high-impact dataset combinations rather than evaluating all possible combinations, thus substantially reducing time cost with minor approximation error. Furthermore, we conduct extensive evaluations on the FL benchmark datasets to demonstrate that our proposed algorithm outperforms a series of representative baselines in terms of efficiency and effectiveness.

cs.LG↗
arXiv2025

FoCTTA: Low-Memory Continual Test-Time Adaptation with Focus

Continual adaptation to domain shifts at test time (CTTA) is crucial for enhancing the intelligence of deep learning enabled IoT applications. However, prevailing TTA methods, which typically update all batch normalization (BN) layers, exhibit two memory inefficiencies. First, the reliance on BN layers for adaptation necessitates large batch sizes, leading to high memory usage. Second, updating all BN layers requires storing the activations of all BN layers for backpropagation, exacerbating the memory demand. Both factors lead to substantial memory costs, making existing solutions impractical for IoT devices. In this paper, we present FoCTTA, a low-memory CTTA strategy. The key is to automatically identify and adapt a few drift-sensitive representation layers, rather than blindly update all BN layers. The shift from BN to representation layers eliminates the need for large batch sizes. Also, by updating adaptation-critical layers only, FoCTTA avoids storing excessive activations. This focused adaptation approach ensures that FoCTTA is not only memory-efficient but also maintains effective adaptation. Evaluations show that FoCTTA improves the adaptation accuracy over the state-of-the-arts by 4.5%, 4.9%, and 14.8% on CIFAR10-C, CIFAR100-C, and ImageNet-C under the same memory constraints. Across various batch sizes, FoCTTA reduces the memory usage by 3-fold on average, while improving the accuracy by 8.1%, 3.6%, and 0.2%, respectively, on the three datasets.

cs.LG↗
arXiv2024

AdaShadow: Responsive Test-time Model Adaptation in Non-stationary Mobile Environments

On-device adapting to continual, unpredictable domain shifts is essential for mobile applications like autonomous driving and augmented reality to deliver seamless user experiences in evolving environments. Test-time adaptation (TTA) emerges as a promising solution by tuning model parameters with unlabeled live data immediately before prediction. However, TTA's unique forward-backward-reforward pipeline notably increases the latency over standard inference, undermining the responsiveness in time-sensitive mobile applications. This paper presents AdaShadow, a responsive test-time adaptation framework for non-stationary mobile data distribution and resource dynamics via selective updates of adaptation-critical layers. Although the tactic is recognized in generic on-device training, TTA's unsupervised and online context presents unique challenges in estimating layer importance and latency, as well as scheduling the optimal layer update plan. AdaShadow addresses these challenges with a backpropagation-free assessor to rapidly identify critical layers, a unit-based runtime predictor to account for resource dynamics in latency estimation, and an online scheduler for prompt layer update planning. Also, AdaShadow incorporates a memory I/O-aware computation reuse scheme to further reduce latency in the reforward pass. Results show that AdaShadow achieves the best accuracy-latency balance under continual shifts. At low memory and energy costs, Adashadow provides a 2x to 3.5x speedup (ms-level) over state-of-the-art TTA methods with comparable accuracy and a 14.8% to 25.4% accuracy boost over efficient supervised methods with similar latency.

cs.LG↗
arXiv2024

Deep Learning Inference on Heterogeneous Mobile Processors: Potentials and Pitfalls

There is a growing demand to deploy computation-intensive deep learning (DL) models on resource-constrained mobile devices for real-time intelligent applications. Equipped with a variety of processing units such as CPUs, GPUs, and NPUs, the mobile devices hold potential to accelerate DL inference via parallel execution across heterogeneous processors. Various efficient parallel methods have been explored to optimize computation distribution, achieve load balance, and minimize communication cost across processors. Yet their practical effectiveness in the dynamic and diverse real-world mobile environment is less explored. This paper presents a holistic empirical study to assess the capabilities and challenges associated with parallel DL inference on heterogeneous mobile processors. Through carefully designed experiments covering various DL models, mobile software/hardware environments, workload patterns, and resource availability, we identify limitations of existing techniques and highlight opportunities for cross-level optimization.

cs.LG↗
arXiv2024

SwapNet: Efficient Swapping for DNN Inference on Edge AI Devices Beyond the Memory Budget

Executing deep neural networks (DNNs) on edge artificial intelligence (AI) devices enables various autonomous mobile computing applications. However, the memory budget of edge AI devices restricts the number and complexity of DNNs allowed in such applications. Existing solutions, such as model compression or cloud offloading, reduce the memory footprint of DNN inference at the cost of decreased model accuracy or autonomy. To avoid these drawbacks, we divide DNN into blocks and swap them in and out in order, such that large DNNs can execute within a small memory budget. Nevertheless, naive swapping on edge AI devices induces significant delays due to the redundant memory operations in the DNN development ecosystem for edge AI devices. To this end, we develop SwapNet, an efficient DNN block swapping middleware for edge AI devices. We systematically eliminate the unnecessary memory operations during block swapping while retaining compatible with the deep learning frameworks, GPU backends, and hardware architectures of edge AI devices. We further showcase the utility of SwapNet via a multi-DNN scheduling scheme. Evaluations on eleven DNN inference tasks in three applications demonstrate that SwapNet achieves almost the same latency as the case with sufficient memory even when DNNs demand 2.32x to 5.81x memory beyond the available budget. The design of SwapNet also provides novel and feasible insights for deploying large language models (LLMs) on edge AI devices in the future.

cs.LG↗
arXiv2024

EchoPFL: Asynchronous Personalized Federated Learning on Mobile Devices with On-Demand Staleness Control

The rise of mobile devices with abundant sensory data and local computing capabilities has driven the trend of federated learning (FL) on these devices. And personalized FL (PFL) emerges to train specific deep models for each mobile device to address data heterogeneity and varying performance preferences. However, mobile training times vary significantly, resulting in either delay (when waiting for slower devices for aggregation) or accuracy decline (when aggregation proceeds without waiting). In response, we propose a shift towards asynchronous PFL, where the server aggregates updates as soon as they are available. Nevertheless, existing asynchronous protocols are unfit for PFL because they are devised for federated training of a single global model. They suffer from slow convergence and decreased accuracy when confronted with severe data heterogeneity prevalent in PFL. Furthermore, they often exclude slower devices for staleness control, which notably compromises accuracy when these devices possess critical personalized data. Therefore, we propose EchoPFL, a coordination mechanism for asynchronous PFL. Central to EchoPFL is to include updates from all mobile devices regardless of their latency. To cope with the inevitable staleness from slow devices, EchoPFL revisits model broadcasting. It intelligently converts the unscalable broadcast to on-demand broadcast, leveraging the asymmetrical bandwidth in wireless networks and the dynamic clustering-based PFL. Experiments show that compared to status quo approaches, EchoPFL achieves a reduction of up to 88.2% in convergence time, an improvement of up to 46% in accuracy, and a decrease of 37% in communication costs

cs.DC↗
arXiv2023

Efficient and Private Federated Trajectory Matching

Federated Trajectory Matching (FTM) is gaining increasing importance in big trajectory data analytics, supporting diverse applications such as public health, law enforcement, and emergency response. FTM retrieves trajectories that match with a query trajectory from a large-scale trajectory database, while safeguarding the privacy of trajectories in both the query and the database. A naive solution to FTM is to process the query through Secure Multi-Party Computation (SMC) across the entire database, which is inherently secure yet inevitably slow due to the massive secure operations. A promising acceleration strategy is to filter irrelevant trajectories from the database based on the query, thus reducing the SMC operations. However, a key challenge is how to publish the query in a way that both preserves privacy and enables efficient trajectory filtering. In this paper, we design GIST, a novel framework for efficient Federated Trajectory Matching. GIST is grounded in Geo-Indistinguishability, a privacy criterion dedicated to locations. It employs a new privacy mechanism for the query that facilitates efficient trajectory filtering. We theoretically prove the privacy guarantee of the mechanism and the accuracy of the filtering strategy of GIST. Extensive evaluations on five real datasets show that GIST is significantly faster and incurs up to 3 orders of magnitude lower communication cost than the state-of-the-arts.

cs.DB↗
arXiv2023

Enabling Resource-efficient AIoT System with Cross-level Optimization: A survey

The emerging field of artificial intelligence of things (AIoT, AI+IoT) is driven by the widespread use of intelligent infrastructures and the impressive success of deep learning (DL). With the deployment of DL on various intelligent infrastructures featuring rich sensors and weak DL computing capabilities, a diverse range of AIoT applications has become possible. However, DL models are notoriously resource-intensive. Existing research strives to realize near-/realtime inference of AIoT live data and low-cost training using AIoT datasets on resource-scare infrastructures. Accordingly, the accuracy and responsiveness of DL models are bounded by resource availability. To this end, the algorithm-system co-design that jointly optimizes the resource-friendly DL models and model-adaptive system scheduling improves the runtime resource availability and thus pushes the performance boundary set by the standalone level. Unlike previous surveys on resource-friendly DL models or hand-crafted DL compilers/frameworks with partially fine-tuned components, this survey aims to provide a broader optimization space for more free resource-performance tradeoffs. The cross-level optimization landscape involves various granularity, including the DL model, computation graph, operator, memory schedule, and hardware instructor in both on-device and distributed paradigms. Furthermore, due to the dynamic nature of AIoT context, which includes heterogeneous hardware, agnostic sensing data, varying user-specified performance demands, and resource constraints, this survey explores the context-aware inter-/intra-device controllers for automatic cross-level adaptation. Additionally, we identify some potential directions for resource-efficient AIoT systems. By consolidating problems and techniques scattered over diverse levels, we aim to help readers understand their connections and stimulate further discussions.

cs.LG↗
arXiv2023

Localised Adaptive Spatial-Temporal Graph Neural Network

Spatial-temporal graph models are prevailing for abstracting and modelling spatial and temporal dependencies. In this work, we ask the following question: whether and to what extent can we localise spatial-temporal graph models? We limit our scope to adaptive spatial-temporal graph neural networks (ASTGNNs), the state-of-the-art model architecture. Our approach to localisation involves sparsifying the spatial graph adjacency matrices. To this end, we propose Adaptive Graph Sparsification (AGS), a graph sparsification algorithm which successfully enables the localisation of ASTGNNs to an extreme extent (fully localisation). We apply AGS to two distinct ASTGNN architectures and nine spatial-temporal datasets. Intriguingly, we observe that spatial graphs in ASTGNNs can be sparsified by over 99.5\% without any decline in test accuracy. Furthermore, even when ASTGNNs are fully localised, becoming graph-less and purely temporal, we record no drop in accuracy for the majority of tested datasets, with only minor accuracy deterioration observed in the remaining datasets. However, when the partially or fully localised ASTGNNs are reinitialised and retrained on the same data, there is a considerable and consistent drop in accuracy. Based on these observations, we reckon that \textit{(i)} in the tested data, the information provided by the spatial dependencies is primarily included in the information provided by the temporal dependencies and, thus, can be essentially ignored for inference; and \textit{(ii)} although the spatial dependencies provide redundant information, it is vital for the effective training of ASTGNNs and thus cannot be ignored during training. Furthermore, the localisation of ASTGNNs holds the potential to reduce the heavy computation overhead required on large-scale spatial-temporal data and further enable the distributed deployment of ASTGNNs.

cs.LG↗
arXiv2023

Towards Capacity-Aware Broker Matching: From Recommendation to Assignment

Online real estate platforms are gaining increasing popularity, where a central issue is to match brokers with clients for potential housing transactions. Mainstream platforms match brokers via top-k recommendation. Yet we observe through extensive data analysis that such top-k recommendation tends to overload the top brokers, which notably degrades their service quality. In this paper, we propose to avoid such overloading in broker matching via the paradigm shift from recommendation to assignment. To this end, we design learned assignment with contextual bandits (LACB), a data-driven capacity-aware assignment scheme for broker matching which estimates broker-specific workload capacity in an online fashion and assigns brokers to clients from a global perspective to maximize the overall service quality. Extensive evaluations on synthetic and real-world datasets from an industrial online real estate platform validate the efficiency and effectiveness of our solution.

cs.DB↗
arXiv2023

POSGen: Personalized Opening Sentence Generation for Online Insurance Sales

The insurance industry is shifting their sales mode from offline to online, in expectation to reach massive potential customers in the digitization era. Due to the complexity and the nature of insurance products, a cost-effective online sales solution is to exploit chatbot AI to raise customers' attention and pass those with interests to human agents for further sales. For high response and conversion rates of customers, it is crucial for the chatbot to initiate a conversation with personalized opening sentences, which are generated with user-specific topic selection and ordering. Such personalized opening sentence generation is challenging because (i) there are limited historical samples for conversation topic recommendation in online insurance sales and (ii) existing text generation schemes often fail to support customized topic ordering based on user preferences. We design POSGen, a personalized opening sentence generation scheme dedicated for online insurance sales. It transfers user embeddings learned from auxiliary online user behaviours to enhance conversation topic recommendation, and exploits a context management unit to arrange the recommended topics in user-specific ordering for opening sentence generation. POSGen is deployed on a real-world online insurance platform. It achieves 2.33x total insurance premium improvement through a two-month global test.

cs.CL↗
arXiv2022

AdaEnlight: Energy-aware Low-light Video Stream Enhancement on Mobile Devices

The ubiquity of camera-embedded devices and the advances in deep learning have stimulated various intelligent mobile video applications. These applications often demand on-device processing of video streams to deliver real-time, high-quality services for privacy and robustness concerns. However, the performance of these applications is constrained by the raw video streams, which tend to be taken with small-aperture cameras of ubiquitous mobile platforms in dim light. Despite extensive low-light video enhancement solutions, they are unfit for deployment to mobile devices due to their complex models and and ignorance of system dynamics like energy budgets. In this paper, we propose AdaEnlight, an energy-aware low-light video stream enhancement system on mobile devices. It achieves real-time video enhancement with competitive visual quality while allowing runtime behavior adaptation to the platform-imposed dynamic energy budgets. We report extensive experiments on diverse datasets, scenarios, and platforms and demonstrate the superiority of AdaEnlight compared with state-of-the-art low-light image and video enhancement solutions.

cs.CV↗