Institution profile

University of Adelaide

Academic institutionaustralasia · au
Official website
Research library579linked papers
Opportunities0open roles
Selected work

Representative Papers

Rethinking the Role of Spatial Mixing

Mar 21, 2025

This work investigates the fundamental role and necessity of spatial mixing operations in visual models. We conduct systematic ablation studies on ResNet and ConvMixer architectures, complemented by PGD-based adversarial evaluation and pixel-wise random shuffling reconstruction tests. Our findings reveal: (1) spatial mixing can be drastically simplified—retaining only random initialization or even full parameter freezing—while preserving over 98% of original ImageNet classification accuracy; (2) such simplification not only preserves performance but substantially enhances adversarial robustness (marked improvement in PGD attack accuracy) and structural recovery capability (successful reconstruction of severely shuffled pixel images). Crucially, we demonstrate for the first time that the core value of spatial mixing lies not in learning complex transformations, but in providing a lightweight, robust, and interpretable implicit structural prior—challenging prevailing assumptions about the necessity of learned spatial aggregation in vision models.

4 citationsRead paper

Maintenance of Structural Hole Spanners in Dynamic Networks

Oct 01, 2021IEEE Conference on Local Computer Networks

To address the low efficiency of maintaining structural hole spanners (SH spanners) in dynamic social networks, this paper proposes the first incremental update framework specifically designed for SH spanners. Unlike static-network approaches that rely on costly full recomputation, our method employs influence-domain analysis to precisely delimit the scope of topological changes and applies a localized re-evaluation strategy—updating only the bridging centrality scores of affected nodes while reusing previously computed results. The core contribution lies in adapting classical static SH spanner algorithms to dynamic settings through a lightweight, efficient incremental mechanism. Extensive experiments on multiple real-world dynamic networks demonstrate that our approach achieves speedups of 3.24×–18.7× over full recomputation, significantly improving maintenance efficiency without compromising identification accuracy.

3 citationsRead paper

Enhancing Close-up Novel View Synthesis via Pseudo-labeling

Mar 20, 2025

Existing NeRF and 3D Gaussian Splatting (3DGS) methods suffer significant performance degradation in close-range novel view synthesis, primarily due to insufficient near-field viewpoint coverage in training data, leading to poor generalization. To address this, we propose a pseudo-label-driven supervised learning framework—introducing the first dedicated benchmark for close-range synthesis. Our approach extends NeRF and 3DGS by integrating self-generated pseudo-labels, multi-view geometric constraints, and targeted data augmentation designed specifically for near-field scenarios. Extensive experiments on our newly established benchmark demonstrate substantial improvements in PSNR and SSIM, markedly enhanced fidelity of fine near-field details, and significantly improved generalization to unseen viewpoints. This work delivers the first systematic solution to close-range novel view synthesis and establishes the first standardized evaluation benchmark for the task.

1 citations1 influentialRead paper

Spectral Conditioning of Attention Improves Transformer Performance

Mar 07, 2026

This work addresses the poor conditioning of the Jacobian matrix in Transformer attention mechanisms, which often leads to training instability and performance degradation. For the first time, it explicitly establishes a theoretical link between the condition number of the attention Jacobian and the spectral properties of the query, key, and value projection matrices. Building on this insight, the paper proposes a general, plug-and-play spectral regularization strategy that improves the Jacobian’s condition number by optimizing the singular value distribution of these projection matrices. Notably, the method requires no architectural modifications and consistently enhances performance across diverse Transformer variants and tasks, demonstrating both its effectiveness and broad applicability.

1 citationsRead paper

DyMA-Fuzz: Dynamic Direct Memory Access Abstraction for Re-hosted Monolithic Firmware Fuzzing

Feb 09, 2026

Existing firmware fuzzing approaches struggle to effectively cover Direct Memory Access (DMA) interfaces in rehosted environments, often missing critical vulnerabilities. This work proposes an automated method that requires neither device datasheets nor manual configuration, leveraging runtime dynamic analysis to infer DMA memory access patterns. It further extends stream-based fuzzing input injection—previously unexplored for DMA—to address challenges such as vendor-specific descriptors, heterogeneous architectures, and variable descriptor locations. Evaluation on 94 firmware samples and eight DMA-related CVEs demonstrates that the approach significantly improves test coverage—by up to 122%—and uncovers previously missed vulnerabilities and execution paths that existing tools fail to detect.

1 citationsRead paper
Recent publications

Latest Papers