Get started
Author

Quanfa Li

Observed institution: Quanzhou Normal University
ORCID: 0009-0002-8784-3555
Observed ORCID: 0009-0002-8784-3555
OpenAlex ID: A5123578027
Layer-dependent dynamic spectral weighting for efficient transformer models
2026 · Zhigao Huang, Yanzhong Yu, et al. · Engineering Applications of Artificial Intelligence · 6
Taming model multiplicity: A unified framework for delay-FPT enumeration of smallest interpretable models
2026 · Zhigao Huang, Shifa Pan, et al. · Information Sciences · 5
Spectral attention for transformers: frequency-domain filtering of attention maps
2026 · Zhigao Huang, Pinghui Wu, et al. · Journal of King Saud University - Computer and Information Sciences · 5
Removing $$\delta $$-dependence in minimal interpretable model learning: distribution conditions and structural parameters
2026 · Zhigao Huang, Shiyan Zheng, et al. · Journal of King Saud University - Computer and Information Sciences · 3
Stability and robustness of minimal majority vote interpretable ensembles
2026 · Quanfa Li, Zhigao Huang, et al. · Scientific Reports · 1
PDF
Accelerating grokking through spectral analysis: A multi-method approach to enhancing abstract rule learning in neural networks
2026 · Zhigao Huang, Shiyan Zheng, et al. · Knowledge-Based Systems · 1
Invariance-Regularized Exact Decision Trees under Cross-Environment Distribution Shift: Code and Data (v2.0.1)
2026 · Zhigao Huang, Pinghui Wu, et al. · Figshare
Frontier-aware variable ordering for OBDD compilation from feature interaction graphs
2026 · Shiyan Zheng, Quanfa Li, et al. · Scientific Reports
PDF
Noise-tolerant optimal decision trees with explicit error budgets for interpretable decision support
2026 · Zhigao Huang, Quanfa Li, et al. · Pattern Recognition
Frequency-Aware Causal Attention for Autoregressive Sequence Modeling
2026 · Zhigao Huang, Miao Pan, et al. · SSRN Electronic Journal
Quantifying Generation Quality for RoPE-Based Long Context Extrapolation
2026 · Mianzhu Peng, Quanfa Li · IEEE Access
Works count: 12
Citation count: 21
H-index: 3
I10-index:
Year
Topic
Type