Institution profile

Macquarie University

Academic institutionaustralasia · au
Official website
Research library489linked papers
Opportunities0open roles
Selected work

Representative Papers

DE3-BERT: Distance-Enhanced Early Exiting for BERT based on Prototypical Networks

Feb 03, 2024arXiv.org

Existing BERT early-exit methods rely solely on per-sample local signals (e.g., entropy) to determine exit decisions, neglecting inter-class global structure—leading to biased reliability estimation and suboptimal exit choices. Method: This work introduces prototype networks into early-exit mechanisms for the first time, proposing a distance-enhanced reliability assessment paradigm that jointly models local entropy and Euclidean distance to class prototypes. A dual-signal, synergistic hybrid gating strategy is designed and integrated into BERT’s hierarchical inference architecture—requiring zero additional parameters or computational overhead. Contribution/Results: The method achieves significant improvements over state-of-the-art approaches on the GLUE benchmark across multiple acceleration ratios, consistently attaining higher accuracy. It exhibits strong generalization across diverse tasks and datasets, while offering enhanced interpretability through geometrically grounded exit decisions based on prototype distances and uncertainty.

6 citations1 influentialRead paper

Information geometry of bosonic Gaussian thermal states

Nov 27, 2024arXiv.org

Characterizing quantum distances between neighboring bosonic Gaussian thermal states and determining fundamental limits on parameter estimation accuracy in their parameter space. Method: We derive, for the first time, closed-form analytical expressions for the symmetric logarithmic derivative (SLD) and its gradient with respect to both the mean vector and the Hamiltonian matrix parametrization of Gaussian thermal states. Based on these, we obtain explicit information matrices for the Fisher–Bures metric and the Kubo–Mori metric. Results: Our framework establishes the quantum Cramér–Rao bound for single-parameter estimation of Gaussian thermal states and provides a rigorous mathematical foundation for natural-gradient optimization on the manifold of Gaussian quantum states. The results directly advance quantum metrology and quantum machine learning, enabling principled design of parameterized quantum algorithms and optimal estimation protocols for continuous-variable systems.

2 citationsRead paper

Federated Learning at the Forefront of Fairness: A Multifaceted Perspective

Aug 01, 2024International Joint Conference on Artificial Intelligence

This work addresses the fairness challenges in federated learning arising from client heterogeneity, which often leads to uneven model performance across participants. To tackle this issue, the paper proposes a systematic taxonomy framework that unifies performance-oriented and capability-oriented fairness strategies, clarifying the technical pathways of existing approaches. Through a comprehensive literature review and taxonomic analysis, the authors construct a structured evaluation metric system for fairness, identifying key challenges and outlining promising directions for future research. This study provides a coherent theoretical foundation, a unified classification perspective, and a forward-looking roadmap to advance fairness-aware federated learning.

2 citationsRead paper

PruneRAG: Confidence-Guided Query Decomposition Trees for Efficient Retrieval-Augmented Generation

Jan 16, 2026

This work addresses the challenges of evidence forgetting and inefficiency in retrieval-augmented generation (RAG) systems during multi-hop reasoning, which stem from unordered query expansion. To mitigate these issues, we propose a confidence-guided query decomposition tree approach that integrates adaptive node expansion, confidence-driven pruning, and fine-grained retrieval anchored at the entity level. This method preserves critical evidence while substantially reducing retrieval overhead. We further introduce evidence forgetting rate as a novel evaluation metric to better assess reasoning fidelity. Experimental results demonstrate that our approach consistently outperforms state-of-the-art methods across multiple multi-hop question answering benchmarks, achieving simultaneous gains in both reasoning accuracy and computational efficiency.

1 citationsRead paper

Majority-Logic Decoding of Binary Locally Recoverable Codes: A Probabilistic Analysis

Jan 13, 2026

This study investigates the error-correction performance of binary linear locally recoverable codes over the binary erasure channel (BEC) and the binary symmetric channel (BSC), addressing a gap in their performance analysis under standard stochastic channel models. Focusing on code constructions with fixed locality and varying availability, the work employs majority-logic decoding to establish, for the first time, explicit upper bounds on both block error rate and bit error rate. Through probabilistic analysis and information-theoretic tools, it reveals a significant gap between worst-case guarantees and typical random-channel performance. Moreover, it proves that under moderately growing availability, the block decoding failure probability vanishes as the code length increases, and the decoder can correct a linear number of errors or erasures with high probability.

1 citationsRead paper
Recent publications

Latest Papers