Yifei Zhu

Chen Tang, Kai Ouyang, Zhi Wang et al.

The exponentially large discrete search space in mixed-precision quantization (MPQ) makes it hard to determine the optimal bit-width for each layer. Previous works usually resort to iterative search methods on the training set, which consume hundreds or even thousands of GPU-hours. In this study, we reveal that some unique learnable parameters in quantization, namely the scale factors in the quantizer, can serve as importance indicators of a layer, reflecting the contribution of that layer to the final accuracy at certain bit-widths. These importance indicators naturally perceive the numerical transformation during quantization-aware training, which can precisely provide quantization sensitivity metrics of layers. However, a deep network always contains hundreds of such indicators, and training them one by one would lead to an excessive time cost. To overcome this issue, we propose a joint training scheme that can obtain all indicators at once. It considerably speeds up the indicators training process by parallelizing the original sequential training processes. With these learned importance indicators, we formulate the MPQ search problem as a one-time integer linear programming (ILP) problem. That avoids the iterative search and significantly reduces search time without limiting the bit-width search space. For example, MPQ search on ResNet18 with our indicators takes only 0.06 s, which improves time efficiency exponentially compared to iterative search methods. Also, extensive experiments show our approach can achieve SOTA accuracy on ImageNet for far-ranging models with various constraints (e.g., BitOps, compress rate). Code is available on https://github.com/1hunters/LIMPQ.

Chen Tang, Haoyu Zhai, Kai Ouyang et al.

Conventional model quantization methods use a fixed quantization scheme to different data samples, which ignores the inherent "recognition difficulty" differences between various samples. We propose to feed different data samples with varying quantization schemes to achieve a data-dependent dynamic inference, at a fine-grained layer level. However, enabling this adaptive inference with changeable layer-wise quantization schemes is challenging because the combination of bit-widths and layers is growing exponentially, making it extremely difficult to train a single model in such a vast searching space and use it in practice. To solve this problem, we present the Arbitrary Bit-width Network (ABN), where the bit-widths of a single deep network can change at runtime for different data samples, with a layer-wise granularity. Specifically, first we build a weight-shared layer-wise quantizable "super-network" in which each layer can be allocated with multiple bit-widths and thus quantized differently on demand. The super-network provides a considerably large number of combinations of bit-widths and layers, each of which can be used during inference without retraining or storing myriad models. Second, based on the well-trained super-network, each layer's runtime bit-width selection decision is modeled as a Markov Decision Process (MDP) and solved by an adaptive inference strategy accordingly. Experiments show that the super-network can be built without accuracy degradation, and the bit-widths allocation of each layer can be adjusted to deal with various inputs on the fly. On ImageNet classification, we achieve 1.1% top1 accuracy improvement while saving 36.2% BitOps.

Qiying Pan, Yifei Zhu

Node embedding aims to map nodes in the complex graph into low-dimensional representations. The real-world large-scale graphs and difficulties of labeling motivate wide studies of unsupervised node embedding problems. Nevertheless, previous effort mostly operates in a centralized setting where a complete graph is given. With the growing awareness of data privacy, data holders who are only aware of one vertex and its neighbours demand greater privacy protection. In this paper, we introduce FedWalk, a random-walk-based unsupervised node embedding algorithm that operates in such a node-level visibility graph with raw graph information remaining locally. FedWalk is designed to offer centralized competitive graph representation capability with data privacy protection and great communication efficiency. FedWalk instantiates the prevalent federated paradigm and contains three modules. We first design a hierarchical clustering tree (HCT) constructor to extract the structural feature of each node. A dynamic time warping algorithm seamlessly handles the structural heterogeneity across different nodes. Based on the constructed HCT, we then design a random walk generator, wherein a sequence encoder is designed to preserve privacy and a two-hop neighbor predictor is designed to save communication cost. The generated random walks are then used to update node embedding based on a SkipGram model. Extensive experiments on two large graphs demonstrate that Fed-Walk achieves competitive representativeness as a centralized node embedding algorithm does with only up to 1.8% Micro-F1 score and 4.4% Marco-F1 score loss while reducing about 6.7 times of inter-device communication per walk.

Qiying Pan, Yifei Zhu, Lingyang Chu

Graph neural networks (GNN) have been widely deployed in real-world networked applications and systems due to their capability to handle graph-structured data. However, the growing awareness of data privacy severely challenges the traditional centralized model training paradigm, where a server holds all the graph information. Federated learning is an emerging collaborative computing paradigm that allows model training without data centralization. Existing federated GNN studies mainly focus on systems where clients hold distinctive graphs or sub-graphs. The practical node-level federated situation, where each client is only aware of its direct neighbors, has yet to be studied. In this paper, we propose the first federated GNN framework called Lumos that supports supervised and unsupervised learning with feature and degree protection on node-level federated graphs. We first design a tree constructor to improve the representation capability given the limited structural information. We further present a Monte Carlo Markov Chain-based algorithm to mitigate the workload imbalance caused by degree heterogeneity with theoretically-guaranteed performance. Based on the constructed tree for each client, a decentralized tree-based GNN trainer is proposed to support versatile training. Extensive experiments demonstrate that Lumos outperforms the baseline with significantly higher accuracy and greatly reduced communication cost and training time.

Shuzhao Xie, Yuan Xue, Yifei Zhu et al.

With the advancement of deep learning techniques, major cloud providers and niche machine learning service providers start to offer their cloud-based machine learning tools, also known as machine learning as a service (MLaaS), to the public. According to our measurement, for the same task, these MLaaSes from different providers have varying performance due to the proprietary datasets, models, etc. Federating different MLaaSes together allows us to improve the analytic performance further. However, naively aggregating results from different MLaaSes not only incurs significant momentary cost but also may lead to sub-optimal performance gain due to the introduction of possible false-positive results. In this paper, we propose Armol, a framework to federate the right selection of MLaaS providers to achieve the best possible analytic performance. We first design a word grouping algorithm to unify the output labels across different providers. We then present a deep combinatorial reinforcement learning based-approach to maximize the accuracy while minimizing the cost. The predictions from the selected providers are then aggregated together using carefully chosen ensemble strategies. The real-world trace-driven evaluation further demonstrates that Armol is able to achieve the same accuracy results with $67\%$ less inference cost.

Yu Liu, Zibo Wang, Yifei Zhu et al.

Federated learning (FL) has emerged as a prevalent distributed machine learning scheme that enables collaborative model training without aggregating raw data. Cloud service providers further embrace Federated Learning as a Service (FLaaS), allowing data analysts to execute their FL training pipelines over differentially-protected data. Due to the intrinsic properties of differential privacy, the enforced privacy level on data blocks can be viewed as a privacy budget that requires careful scheduling to cater to diverse training pipelines. Existing privacy budget scheduling studies prioritize either efficiency or fairness individually. In this paper, we propose DPBalance, a novel privacy budget scheduling mechanism that jointly optimizes both efficiency and fairness. We first develop a comprehensive utility function incorporating data analyst-level dominant shares and FL-specific performance metrics. A sequential allocation mechanism is then designed using the Lagrange multiplier method and effective greedy heuristics. We theoretically prove that DPBalance satisfies Pareto Efficiency, Sharing Incentive, Envy-Freeness, and Weak Strategy Proofness. We also theoretically prove the existence of a fairness-efficiency tradeoff in privacy budgeting. Extensive experiments demonstrate that DPBalance outperforms state-of-the-art solutions, achieving an average efficiency improvement of $1.44\times \sim 3.49 \times$, and an average fairness improvement of $1.37\times \sim 24.32 \times$.

Zhe Wang, Jinghang Li, Yifei Zhu

Free-viewpoint video (FVV) enables immersive viewing experiences by allowing users to view scenes from arbitrary perspectives. As a prominent reconstruction technique for FVV generation, 4D Gaussian Splatting (4DGS) models dynamic scenes with time-varying 3D Gaussian ellipsoids and achieves high-quality rendering via fast rasterization. However, existing 4DGS approaches suffer from quality degradation over long sequences and impose substantial bandwidth and storage overhead, limiting their applicability in real-time and wide-scale deployments. Therefore, we present AirGS, a streaming-optimized 4DGS framework that rearchitects the training and delivery pipeline to enable high-quality, low-latency FVV experiences. AirGS converts Gaussian video streams into multi-channel 2D formats and intelligently identifies keyframes to enhance frame reconstruction quality. It further combines temporal coherence with inflation loss to reduce training time and representation size. To support communication-efficient transmission, AirGS models 4DGS delivery as an integer linear programming problem and design a lightweight pruning level selection algorithm to adaptively prune the Gaussian updates to be transmitted, balancing reconstruction quality and bandwidth consumption. Extensive experiments demonstrate that AirGS reduces quality deviation in PSNR by more than 20% when scene changes, maintains frame-level PSNR consistently above 30, accelerates training by 6 times, reduces per-frame transmission size by nearly 50% compared to the SOTA 4DGS approaches.

Haichao Ji, Zibo Wang, Cheng Pan et al.

Large Language Models (LLMs) have shown great promise in automating data analytics tasks by interpreting natural language queries and generating multi-operation execution plans. However, existing LLM-agent-based analytics frameworks operate under the assumption of centralized data access, offering little to no privacy protection. In contrast, federated analytics (FA) enables privacy-preserving computation across distributed data sources, but lacks support for natural language input and requires structured, machine-readable queries. In this work, we present LAFA, the first system that integrates LLM-agent-based data analytics with FA. LAFA introduces a hierarchical multi-agent architecture that accepts natural language queries and transforms them into optimized, executable FA workflows. A coarse-grained planner first decomposes complex queries into sub-queries, while a fine-grained planner maps each subquery into a Directed Acyclic Graph of FA operations using prior structural knowledge. To improve execution efficiency, an optimizer agent rewrites and merges multiple DAGs, eliminating redundant operations and minimizing computational and communicational overhead. Our experiments demonstrate that LAFA consistently outperforms baseline prompting strategies by achieving higher execution plan success rates and reducing resource-intensive FA operations by a substantial margin. This work establishes a practical foundation for privacy-preserving, LLM-driven analytics that supports natural language input in the FA setting.

Zhe Wang, Yifei Zhu

Neural Radiance Fields (NeRF) is an emerging technique to synthesize 3D objects from 2D images with a wide range of potential applications. However, rendering existing NeRF models is extremely computation intensive, making it challenging to support real-time interaction on mobile devices. In this paper, we take the first initiative to examine the state-of-the-art real-time NeRF rendering technique from a system perspective. We first define the entire working pipeline of the NeRF serving system. We then identify possible control knobs that are critical to the system from the communication, computation, and visual performance perspective. Furthermore, an extensive measurement study is conducted to reveal the effects of these control knobs on system performance. Our measurement results reveal that different control knobs contribute differently towards improving the system performance, with the mesh granularity being the most effective knob and the quantization being the least effective knob. In addition, diverse hardware device settings and network conditions have to be considered to fully unleash the benefit of operating under the appropriate knobs