arXiv subjects

Weiqiang Sun

Publications and source records attributed to Weiqiang Sun.

9 recordsLinked to original sources

arXiv2026

Switching Efficiency: A Novel Framework for Dissecting AI Data Center Network Efficiency

Communication is pivotal in LLM training, and a thorough analysis of the communication efficiency of AI data center (AIDC) network is essential for guiding the design of these capital-intensive clusters. However, conventional metrics are inadequate for such analysis, as they do not directly link network activity to computational progress and lack granularity to diagnose the impact of different network design patterns. To address this, we introduce a metric framework, the Switching Efficiency Framework, whose core metric - Switching Efficiency ($\eta$) - quantifies computationally effective data throughput per unit switching capacity. We further decompose $\eta$ into three factors - Data, Routing Efficiency, and Port Utilization to facilitate analysis of distinct communication bottlenecks. Using this metric framework, we demonstrate how the symmetric, distributed switching of 3D-Torus and the centralized, hierarchical switching of Rail-Optimized architecture align with sparse or imbalanced LLM training traffic, and show that All-to-All traffic from Mixture-of-Experts models severely degrades their port utilization and routing efficiency. Our analysis also demonstrates how key design choices - such as adjusting switching resource allocation, expanding server size, adopting in-network computing, and multi-plane design - positively influence distinct facets of communication efficiency. Ultimately, the Switching Efficiency Framework provides an analytical tool for analyzing efficiency bottlenecks, thereby informing the design of future-generation AIDC networks.

cs.NI↗

arXiv2026

Beyond Traffic Matrix: DELTA -- A DAG-Aware OCS Logical Topology Optimization for AIDCs

The rapid scaling of large language models (LLMs) exacerbates communication bottlenecks in AI data centers (AIDCs). To overcome this, optical circuit switches (OCS) are increasingly adopted for their superior bandwidth capacity and energy efficiency. However, their reconfiguration overhead precludes intra-iteration topology update, necessitating a priori engineering of a static topology to absorb time-varying LLM traffic. Existing methods engineer these topologies based on traffic matrices. However, this representation obscures the bursty concurrent bandwidth demands dictated by parallelization strategies and fails to account for the independent channels required for concurrent communication. To address this, we propose DELTA, an efficient logical topology optimization framework for AIDCs that leverages the computation-communication directed acyclic graph (DAG) to encode time-varying traffic patterns into a Mixed-Integer Linear Programming (MILP) model, while exploiting the temporal slack of non-critical tasks to save optical ports without penalizing iteration makespan. By pioneering a variable-length time interval formulation, DELTA significantly reduces the solution space compared to the fixed-time-step formulation. To scale to thousand-GPU clusters, we design a dual-track acceleration strategy that combines search space pruning (reducing complexity from quadratic to linear) with heuristic hot-starting. Evaluations on large-scale LLM workloads show that DELTA reduces communication time by up to 17.5% compared to state-of-the-art traffic-matrix-based baselines. Furthermore, the framework reduces optical port consumption by at least 20%; dynamically reallocating these surplus ports to bandwidth-bottlenecked workloads reduces their performance gap relative to ideal non-blocking electrical networks by up to 26.1%, ultimately enabling most workloads to achieve near-ideal performance.

cs.NI↗

arXiv2025

Why Attention Fails: The Degeneration of Transformers into MLPs in Time Series Forecasting

Transformer-based architectures achieved high performance in natural language processing and computer vision, yet many studies have shown that they have not demonstrated a clear advantage in time series forecasting and even underperform simple linear baselines in some cases. However, most of these studies have not thoroughly explored the reasons behind the failure of transformers. To better understand time-series transformers(TST), we designed a series of experiments, progressively modifying transformers into MLPs to investigate the impact of the attention mechanism. Surprisingly, transformer blocks often degenerate into simple MLPs in existing time-series transformers. We designed a interpretable dataset to investigate the reasons behind the failure of the attention mechanism and revealed that the attention mechanism is not working in the expected way. We theoretically analyzed the reasons behind this phenomenon, demonstrating that the current embedding methods fail to allow transformers to function in a well-structured latent space, and further analyzed the deeper underlying causes of the failure of embedding.

cs.LG↗

arXiv2025

TBGRecall: A Generative Retrieval Model for E-commerce Recommendation Scenarios

Recommendation systems are essential tools in modern e-commerce, facilitating personalized user experiences by suggesting relevant products. Recent advancements in generative models have demonstrated potential in enhancing recommendation systems; however, these models often exhibit limitations in optimizing retrieval tasks, primarily due to their reliance on autoregressive generation mechanisms. Conventional approaches introduce sequential dependencies that impede efficient retrieval, as they are inherently unsuitable for generating multiple items without positional constraints within a single request session. To address these limitations, we propose TBGRecall, a framework integrating Next Session Prediction (NSP), designed to enhance generative retrieval models for e-commerce applications. Our framework reformulation involves partitioning input samples into multi-session sequences, where each sequence comprises a session token followed by a set of item tokens, and then further incorporate multiple optimizations tailored to the generative task in retrieval scenarios. In terms of training methodology, our pipeline integrates limited historical data pre-training with stochastic partial incremental training, significantly improving training efficiency and emphasizing the superiority of data recency over sheer data volume. Our extensive experiments, conducted on public benchmarks alongside a large-scale industrial dataset from TaoBao, show TBGRecall outperforms the state-of-the-art recommendation methods, and exhibits a clear scaling law trend. Ultimately, NSP represents a significant advancement in the effectiveness of generative recommendation systems for e-commerce applications.

cs.IR↗

arXiv2025

Efficient Decomposition of Forman-Ricci Curvature on Vietoris-Rips Complexes and Data Applications

Discrete Forman-Ricci curvature (FRC) is an efficient tool that characterizes essential geometrical features and associated transitions of real-world networks, extending seamlessly to higher-dimensional computations in simplicial complexes. In this article, we provide two major advancements: First, we give a decomposition for FRC that inhently allows a local computations of FRC. Second, we construct a set-theoretical proof enabling an efficient algorithms for the local computation of FRC in Vietoris-Rips (VR) complexes. Our findings open new avenues for geometric computations in VR complexes and highlight an essential yet under-explored aspect of data analysis and visualisation: the geometry underpinning statistical patterns.

math.GT↗

arXiv2023

Twotier -- A Layered Analysis of Backbone Members in a Moderate Sized Community Sports Organization

Backbone members are recognized as essential parts of an organization, yet their role and mechanisms of functioning in networks are not fully understood. In this paper, we propose a new framework called Twotier to analyze the evolution of community sports organizations (CSOs) and the role of backbone members. Tier-one establishes a dynamic user interaction network based on grouping relationships, and weighted k-shell decomposition is used to select backbone members. We perform community detection and capture the evolution of two separate sub-networks: one formed by backbone members and the other formed by other members. In Tier-two, the sub-networks are abstracted, revealing a core-periphery structure in the organization where backbone members serve as bridges connecting all parts of the network. Our findings suggest that relying on backbone members can keep newcomers actively involved in rewarding activities, while non-rewarding activities solidify relations between backbone members.

cs.SI↗

arXiv2023

Understanding the power-law nature of participation in community sports organizations

The improvement of living standards and awareness of chronic diseases have increased the importance of community sports organizations in promoting the physical activity levels of the public. However, limited understanding of human behavior in this context often leads to suboptimal resource utilization. In this study, we analyzed the participation behavior of 2,956 members with a time span of 6 years in a community sports organization. Our study reveals that, at the population level, the participation frequency in activities adheres to a power-law distribution. To understand the underlying mechanisms driving crowd participation, we introduce a novel behavioral model called HFBI (Habit-Formation and Behavioral Inertia), demonstrating a robust fit to the observed power-law distribution. The habit formation mechanism indicates that individuals who are more engaged are more likely to maintain participation, while the behavioral inertia mechanism suggests that individuals' willingness to participate in activities diminishes with their absences from activities. At the individual level, our analysis reveals a burst-quiet participation pattern, with bursts often commencing with incentive activities. We also find a power-law distribution in the intervals between individual participations. Our research offers valuable insights into the complex dynamics of human participation in community sports activity and provides a theoretical foundation to inform intervention design. Furthermore, the flexibility of our model enables its application to other data exhibiting power-law properties, broadening its potential impact beyond the realm of community sports.

cs.SI↗

arXiv2021

Computation Rate Maximum for Mobile Terminals in UAV-assisted Wireless Powered MEC Networks with Fairness Constraint

This paper investigates an unmanned aerial vehicle (UAV)-assisted wireless powered mobile-edge computing (MEC) system, where the UAV powers the mobile terminals by wireless power transfer (WPT) and provides computation service for them. We aim to maximize the computation rate of terminals while ensuring fairness among them. Considering the random trajectories of mobile terminals, we propose a soft actor-critic (SAC)-based UAV trajectory planning and resource allocation (SAC-TR) algorithm, which combines off-policy and maximum entropy reinforcement learning to promote the convergence of the algorithm. We design the reward as a heterogeneous function of computation rate, fairness, and reaching of destination. Simulation results show that SAC-TR can quickly adapt to varying network environments and outperform representative benchmarks in a variety of situations.

cs.NI↗

arXiv2020

Cost of Dietary Data Acquisition with Smart Group Catering

The need for dietary data management is growing with public awareness of food intakes. As a result, there are increasing deployments of smart canteens where dietary data is collected through either Radio Frequency Identification (RFID) or Computer Vision(CV)-based solutions. As human labor is involved in both cases, manpower allocation is critical to data quality. Where manpower requirements are underestimated, data quality is compromised. This paper has studied the relation between the quality of dietary data and the manpower invested, using numerical simulations based on real data collected from multiple smart canteens. We found that in both RFID and CV-based systems, the long-term cost of dietary data acquisition is dominated by manpower. Our study provides a comprehensive understanding of the cost composition for dietary data acquisition and useful insights toward future cost effective systems.

cs.CY↗