jiaqing lyu

jiaqing lyu


Research Focus & Metrics

Academic Career Length

1Year (2026-Present)

(Calculated once Scilit publication data is synced: current year − year of first publication + 1)

Representative Publications

View All Publications
21 August 2026 in Preprints.org
Verified
Preprint
Open access
PDF file
KBKangyou Bao
Kangyou Bao
WGWenqi Gu
Wenqi Gu
JLjiaqing lyu
jiaqing lyu
XDXizheng Deng
Xizheng Deng
Carlo Vittorio CannistraciCarlo Vittorio Cannistraci
Carlo Vittorio Cannistraci
ANN-to-SNN conversion is an important approach for obtaining high-performance spiking neural networks (SNNs), yet current conversion methodologies predominantly focus on dense architectures, overlooking the struc- tural sparsity that characterizes biological neural systems. In this work, we propose a sparse ANN-to-SNN conversion framework that inherits learned sparse topology into SNNs. We instantiate this framework with Cannistraci-Hebb Training (CHT), a brain-inspired dynamic sparse training method. Across CNN experiments on CIFAR-10, CIFAR-100, and DVSGesture, and ViT experiments on ImageNet-1K, SNNs produced by our framework achieve near-dense performance while outperforming SNNs obtained from static pruning and sparse- training baselines in accuracy. Compared with dense SNN counterparts in the principal CIFAR and ImageNet comparisons, our framework also substantially reduces energy consumption. Beyond these accuracy-efficiency gains, the resulting sparse SNNs are characterized by biologically plausible structural properties, including meta- depth and scale-free organization. These results suggest that learned sparse topology is a useful design dimension for ANN-to-SNN conversion, enabling high-accuracy sparse SNNs with substantially reduced energy consumption.

13 May 2026 in Preprints.org
Verified
Preprint
Open access
PDF file
JLjiaqing lyu
jiaqing lyu
MWMichael Wirz
Michael Wirz
Carlo Vittorio CannistraciCarlo Vittorio Cannistraci
Carlo Vittorio Cannistraci
Early 2026 has witnessed significant volatility in the oil market, and an energy crisis is expected in the coming months. Large-scale LLM inference continues to consume substantial power in data centers and improving inference efficiency is therefore increasingly important for energy sustainability of AI economy. Semi-structured N:M sparsity, most notably 2:4 (50%) and potentially 1:4 (75%), offers a hardware-friendly path to lower compute and energy, and has been supported in modern GPU designs. Yet existing training methods for 2:4 sparsity (e.g., STE-based approaches) often incur large accuracy drops relative to dense baselines, and practical support for 1:4 remains limited in current software stacks. As a result, attention has shifted toward quantization and mixture-of-experts, leaving high-sparsity N:M pre-training underexplored. Here we introduce a paradigm shift: we treat neural networks as complex systems whose sparse connectivity can be trained using network-science principles formalized by Cannistraci–Hebb sparse-to-sparse training (CHT), coupled with a tailored optimizer. We propose CHTsNM, a sparse-to-sparse training framework centered on Topology-Aware Newton–Schulz (TANS) optimization. TANS makes Newton–Schulz-style matrix updates compatible with dynamically changing semi-structured sparse topologies via active-mask projection, active-support RMS matching, and refresh-aware ramping after topology updates. CHTsNM further incorporates two lightweight mechanisms: Contextually Modulated LoRA (CoMoLoRA) for input-adaptive low-rank residual compensation, and Motif Pattern Revisitation (MPR) to improve exploration of legal row-wise N:M patterns. Across 4 LLaMA pre-training benchmarks, CHTsNM with 2:4 sparsity achieves performance close to dense baselines on most tasks and yields sparse-over-dense gains on 8 tasks. 1:4 sparsity approaches dense performance, though does not yet consistently surpass it. For hardware evaluation, we report measured speedups for native 2:4 execution on current NVIDIA GPUs, and provide a clearly labeled CSR sparse-GEMM surrogate analysis to estimate the acceleration potential of 1:4. Overall, although not implement on hardware yet, our results identify 1:4 sparse pre-training as a promising direction and establish TANS sparse-to-sparse optimization as a practical step toward future high-sparsity N:M accelerators.