Selected work

Representative Papers

Counterfactual Explainability Framework With CycleGAN And Counterfactual-Classifier Alignnment Score for Retinal Disease Classification

Jul 23, 2026

Current deep learning models for retinal disease classification suffer from limited interpretability and an inability to reliably link predictions to clinically relevant lesion regions, hindering their clinical deployment. To address this, this work proposes CounterFundus, a novel framework that leverages CycleGAN to generate healthy counterfactual images corresponding to pathological fundus images. Lesion localization is achieved through difference maps between original and counterfactual images. The study introduces CCAS—a unified evaluation metric combining Spearman correlation, Intersection over Union (IoU), and pointing accuracy—to quantitatively assess the spatial alignment between counterfactual explanations and classifier saliency maps. Experiments demonstrate that the generated counterfactuals exhibit strong alignment with classification evidence across all CCAS dimensions, and that CCAS-guided counterfactual data augmentation significantly enhances downstream classification performance.

0 citationsRead paper

Beyond Self-Attention: Sub-Quadratic Vision Transformers for Fast Image Captioning

Jun 07, 2026

This work addresses the high computational complexity and inadequate local feature modeling of vision Transformers in image captioning by proposing a sub-quadratic visual Transformer architecture based on Gaussian Mixture Model (GMM) soft clustering. Instead of conventional self-attention, the method employs an Expectation-Maximization (EM) algorithm to perform semantics-aware soft clustering of image patches, enabling linear-complexity feature aggregation. The resulting visual representations are fed into a GPT-based autoregressive decoder to generate captions. Experimental results on the Flickr30K dataset demonstrate that the proposed model significantly reduces computational overhead while maintaining or even enhancing semantic expressiveness, achieving a dual improvement in both efficiency and performance.

0 citationsRead paper

SKG-Eval: Stateful Evaluation of Multi-Turn Dialogue via Incremental Semantic Knowledge Graphs

May 15, 2026

Existing automatic evaluation methods struggle to effectively detect long-range inconsistencies in multi-turn dialogues, such as contradictions, topic drift, and entity conflicts. This work proposes the first stateful evaluation framework based on an incremental Semantic Knowledge Graph (SKG), which dynamically tracks entities, relations, and commitments throughout a dialogue by leveraging structured triple extraction and graph embeddings. The approach integrates three complementary signals—local relevance, historical consistency, and logical coherence—to explicitly identify cross-turn contradictions without relying on large language models or natural language inference systems, thereby producing auditable evaluation certificates. Experimental results demonstrate that the framework significantly improves correlation with human judgments across multiple benchmarks, substantially enhances detection of long-range inconsistencies, and yields deterministic scores under fixed inputs.

0 citationsRead paper

An Ideal Random Number Generator Based on Quantum Fluctuations and Rotating Wheel for Secure Image Encryption

Mar 13, 2026

This study addresses the insufficient randomness of cryptographic keys in secure digital image transmission by proposing a true random number generation method that integrates quantum fluctuations with an algorithm-driven rotary wheel mechanism. For the first time, the quantum kicked rotor model is combined with a mechanical rotary wheel, leveraging time-varying rotation speeds and multi-point sampling to dynamically produce high-entropy sequences, thereby significantly enhancing unpredictability. The generated sequences achieve an entropy value of 7.997, approaching the theoretical maximum of 8. When applied to image encryption, the method yields an NPCR of 99.60%, near-zero pixel correlation, and low PSNR, demonstrating its suitability for high-security consumer applications such as mobile healthcare and biometric authentication.

0 citationsRead paper

Subgroup perfect codes of $ S_n $ in Cayley graphs

Jan 31, 2026

This study investigates the characterization of subgroups of the symmetric group $S_n$ that serve as perfect codes in Cayley graphs, with a focus on cyclic 2-subgroups. By integrating techniques from group theory, graph theory, and combinatorial coding theory, the work systematically explores the relationship between the algebraic structure of subgroups and their coding properties in Cayley graphs. It provides the first complete classification of cyclic 2-subgroups of $S_n$ that can be realized as perfect codes, covering both abelian and non-abelian cases, and extends these results to broader classes of subgroups. The findings yield precise structural characterizations supported by numerous explicit examples, thereby advancing the interplay between algebraic graph theory and coding theory.

0 citationsRead paper

Recent publications

Latest Papers

Counterfactual Explainability Framework With CycleGAN And Counterfactual-Classifier Alignnment Score for Retinal Disease Classification

Jul 23, 2026

Current deep learning models for retinal disease classification suffer from limited interpretability and an inability to reliably link predictions to clinically relevant lesion regions, hindering their clinical deployment. To address this, this work proposes CounterFundus, a novel framework that leverages CycleGAN to generate healthy counterfactual images corresponding to pathological fundus images. Lesion localization is achieved through difference maps between original and counterfactual images. The study introduces CCAS—a unified evaluation metric combining Spearman correlation, Intersection over Union (IoU), and pointing accuracy—to quantitatively assess the spatial alignment between counterfactual explanations and classifier saliency maps. Experiments demonstrate that the generated counterfactuals exhibit strong alignment with classification evidence across all CCAS dimensions, and that CCAS-guided counterfactual data augmentation significantly enhances downstream classification performance.

0 citationsRead paper

Beyond Self-Attention: Sub-Quadratic Vision Transformers for Fast Image Captioning

Jun 07, 2026

This work addresses the high computational complexity and inadequate local feature modeling of vision Transformers in image captioning by proposing a sub-quadratic visual Transformer architecture based on Gaussian Mixture Model (GMM) soft clustering. Instead of conventional self-attention, the method employs an Expectation-Maximization (EM) algorithm to perform semantics-aware soft clustering of image patches, enabling linear-complexity feature aggregation. The resulting visual representations are fed into a GPT-based autoregressive decoder to generate captions. Experimental results on the Flickr30K dataset demonstrate that the proposed model significantly reduces computational overhead while maintaining or even enhancing semantic expressiveness, achieving a dual improvement in both efficiency and performance.

0 citationsRead paper

SKG-Eval: Stateful Evaluation of Multi-Turn Dialogue via Incremental Semantic Knowledge Graphs

May 15, 2026

Existing automatic evaluation methods struggle to effectively detect long-range inconsistencies in multi-turn dialogues, such as contradictions, topic drift, and entity conflicts. This work proposes the first stateful evaluation framework based on an incremental Semantic Knowledge Graph (SKG), which dynamically tracks entities, relations, and commitments throughout a dialogue by leveraging structured triple extraction and graph embeddings. The approach integrates three complementary signals—local relevance, historical consistency, and logical coherence—to explicitly identify cross-turn contradictions without relying on large language models or natural language inference systems, thereby producing auditable evaluation certificates. Experimental results demonstrate that the framework significantly improves correlation with human judgments across multiple benchmarks, substantially enhances detection of long-range inconsistencies, and yields deterministic scores under fixed inputs.

0 citationsRead paper

An Ideal Random Number Generator Based on Quantum Fluctuations and Rotating Wheel for Secure Image Encryption

Mar 13, 2026

This study addresses the insufficient randomness of cryptographic keys in secure digital image transmission by proposing a true random number generation method that integrates quantum fluctuations with an algorithm-driven rotary wheel mechanism. For the first time, the quantum kicked rotor model is combined with a mechanical rotary wheel, leveraging time-varying rotation speeds and multi-point sampling to dynamically produce high-entropy sequences, thereby significantly enhancing unpredictability. The generated sequences achieve an entropy value of 7.997, approaching the theoretical maximum of 8. When applied to image encryption, the method yields an NPCR of 99.60%, near-zero pixel correlation, and low PSNR, demonstrating its suitability for high-security consumer applications such as mobile healthcare and biometric authentication.

0 citationsRead paper

Subgroup perfect codes of $ S_n $ in Cayley graphs

Jan 31, 2026

This study investigates the characterization of subgroups of the symmetric group $S_n$ that serve as perfect codes in Cayley graphs, with a focus on cyclic 2-subgroups. By integrating techniques from group theory, graph theory, and combinatorial coding theory, the work systematically explores the relationship between the algebraic structure of subgroups and their coding properties in Cayley graphs. It provides the first complete classification of cyclic 2-subgroups of $S_n$ that can be realized as perfect codes, covering both abelian and non-abelian cases, and extends these results to broader classes of subgroups. The findings yield precise structural characterizations supported by numerous explicit examples, thereby advancing the interplay between algebraic graph theory and coding theory.

0 citationsRead paper