Selected work

Representative Papers

Unconditional Time and Space Complexity Lower Bounds for Intersection Non-Emptiness

Nov 28, 2025

This paper investigates the computational complexity of the DFA intersection non-emptiness problem. We establish the first unconditional time lower bound of Ω(n²/log³n loglog²n), breaking prior conditional lower bounds that relied on unproven hypotheses, and derive tight space lower bounds. Technically, our approach combines nondeterministic logspace reductions with Williams’ (2025) deterministic time–space-efficient simulation framework, while strengthening the intrinsic connection between time–space trade-offs in simulation. Our main contributions are: (1) the first unconditional quadratic-time lower bound for this problem; and (2) a structural result showing that if DFA intersection non-emptiness is not solvable in fixed-polynomial time, then major complexity class collapses follow—including PTIME ⊆ DSPACE(nᶜ) for some constant c and PSPACE = EXPTIME—thereby deepening the foundational links between automata theory and central complexity classes (P, PSPACE, EXPTIME).

0 citationsRead paper

Biases in Large Language Model-Elicited Text: A Case Study in Natural Language Inference

Mar 06, 2025International Conference on Computational Linguistics

This work investigates whether natural language inference (NLI) data generated by large language models (LLMs)—specifically GPT-4, Llama-2-70b, and Mistral-7b—inherit annotation artifacts and societal biases (e.g., gender, race, age) present in human-annotated NLI datasets. We construct LLM-generated NLI subsets and empirically identify severe hypothesis exclusivity bias and stereotypical social biases—first such evidence in synthetic NLI data. To detect these biases systematically, we propose a dual-path framework: (1) a fine-tuned BERT-based hypothesis exclusivity classifier, and (2) pointwise mutual information (PMI) analysis for bias-associated lexical patterns. Experiments show the framework achieves 86–96% classification accuracy on LLM-generated data—significantly outperforming its performance on human-annotated data—and quantitatively identifies multiple bias-correlated lexical terms. Our findings provide both novel methodology and critical empirical evidence for bias assessment and mitigation in LLM-synthesized training data.

0 citationsRead paper

Recent publications

Latest Papers

Unconditional Time and Space Complexity Lower Bounds for Intersection Non-Emptiness

Nov 28, 2025

This paper investigates the computational complexity of the DFA intersection non-emptiness problem. We establish the first unconditional time lower bound of Ω(n²/log³n loglog²n), breaking prior conditional lower bounds that relied on unproven hypotheses, and derive tight space lower bounds. Technically, our approach combines nondeterministic logspace reductions with Williams’ (2025) deterministic time–space-efficient simulation framework, while strengthening the intrinsic connection between time–space trade-offs in simulation. Our main contributions are: (1) the first unconditional quadratic-time lower bound for this problem; and (2) a structural result showing that if DFA intersection non-emptiness is not solvable in fixed-polynomial time, then major complexity class collapses follow—including PTIME ⊆ DSPACE(nᶜ) for some constant c and PSPACE = EXPTIME—thereby deepening the foundational links between automata theory and central complexity classes (P, PSPACE, EXPTIME).

0 citationsRead paper

Biases in Large Language Model-Elicited Text: A Case Study in Natural Language Inference

Mar 06, 2025International Conference on Computational Linguistics

This work investigates whether natural language inference (NLI) data generated by large language models (LLMs)—specifically GPT-4, Llama-2-70b, and Mistral-7b—inherit annotation artifacts and societal biases (e.g., gender, race, age) present in human-annotated NLI datasets. We construct LLM-generated NLI subsets and empirically identify severe hypothesis exclusivity bias and stereotypical social biases—first such evidence in synthetic NLI data. To detect these biases systematically, we propose a dual-path framework: (1) a fine-tuned BERT-based hypothesis exclusivity classifier, and (2) pointwise mutual information (PMI) analysis for bias-associated lexical patterns. Experiments show the framework achieves 86–96% classification accuracy on LLM-generated data—significantly outperforming its performance on human-annotated data—and quantitatively identifies multiple bias-correlated lexical terms. Our findings provide both novel methodology and critical empirical evidence for bias assessment and mitigation in LLM-synthesized training data.

0 citationsRead paper