Selected work

Representative Papers

LLM-Guided Program Evolution for Targeted Black-Box Attacks on Perceptual Hash Algorithms

Jul 13, 2026

This work addresses the lack of provable robustness in perceptual hashing algorithms under adversarial perturbations by introducing GigaEvo/OpenEvolve, a large language model (LLM)-guided program evolution framework. For the first time, LLM-driven program synthesis is applied to black-box attacks on perceptual hashing, circumventing the need for internal algorithm access and effectively handling the discrete and non-differentiable nature of hash outputs. By optimizing a composite scoring metric, the method achieves high-efficiency attacks with minimal perturbation. Experimental results demonstrate that the approach significantly outperforms existing black-box attacks on pHash, PDQ, PhotoDNA, and NeuralHash, requiring fewer queries, achieving lower L2 distortion, and reducing the composite attack score by up to 41.2%, thereby exposing previously undisclosed security vulnerabilities in mainstream content moderation systems.

0 citationsRead paper

A Foundation Model for Multimodal Event Sequences in Financial Applications

Jul 10, 2026

This work addresses the limitations of traditional financial forecasting approaches, which rely on task-specific models and handcrafted features, struggle to integrate heterogeneous data (e.g., transactional and digital interaction logs), and exhibit poor reusability. The authors propose the first multimodal event-sequence foundation model tailored for financial applications. It unifies users’ multisource behaviors into temporal sequences and leverages a Transformer architecture pretrained via next-event prediction to learn general-purpose representations. These representations, combined with existing features, enable lightweight fine-tuning across diverse downstream tasks. The model achieves early fusion of multimodal events and cross-task representation sharing, effectively breaking down task silos. Deployed at a major bank in Eastern Europe, it significantly outperforms conventional models, enhancing key business metrics while reducing development costs.

0 citationsRead paper

A Stationary-Distribution Theory for Triplet-Based Plateau Search in Random Forest Ensemble-Size Selection

Jun 29, 2026

This work addresses the trade-off between computational cost and model stability in selecting the number of trees in random forests, noting that existing plateau-based tuning methods exhibit nondeterministic fluctuations even after apparent convergence. The study formulates ensemble size selection as a birth–death Markov chain on a geometric grid and derives its stationary distribution using local balance principles, revealing that plateau search is inherently a stochastic process. By introducing a symmetric correction update rule, the authors establish a theoretical framework for the stationary distribution of ternary plateau search, analyzed via a folded normal approximation centered at the optimum and a local Gaussian approximation. They prove that both the stationary center \(B_*\) and its standard deviation \(\sigma_{B,*}\) scale as \(O(\varepsilon^{-2})\), the variance as \(O(\varepsilon^{-4})\), and crucially, that the relative width of the distribution depends solely on the scaling factor and update rule, independent of the precision parameter \(\varepsilon\).

0 citationsRead paper

How Many Trees in a Random Forest? A Revisited Approach with Plateau Search and Optuna Integration

Jun 02, 2026

This work addresses the challenge of optimizing the number of trees in random forests, a hyperparameter whose performance typically improves monotonically with ensemble size, thereby confounding conventional hyperparameter optimization methods due to sensitivity to preset bounds or early-stopping noise. The authors propose a triplet plateau search algorithm that excludes this parameter from the Tree-structured Parzen Estimator (TPE) search space and instead employs a sliding window to monitor relative changes in out-of-bag (OOB) scores across three consecutive ensemble sizes. This approach adaptively identifies the minimal ensemble size satisfying a target performance criterion, eliminating the need for a predefined upper bound while offering both automation and interpretability. Theoretical analysis links the plateau criterion to the asymptotic performance gap and variance estimation. Empirical results demonstrate substantial reductions in tree count on most datasets, while automatically selecting larger ensembles on high-dimensional biological benchmarks such as Arcene and Dorothea, confirming the method’s adaptivity and effectiveness.

0 citationsRead paper

PersonalAI 2.0: Enhancing knowledge graph traversal/retrieval with planning mechanism for Personalized LLM Agents

May 13, 2026

Existing Graph Retrieval-Augmented Generation (GraphRAG) approaches for personalized large language model (LLM) agents suffer from static retrieval and a lack of planning mechanisms, leading to low factual accuracy and increased hallucination in question answering. To address these limitations, this work proposes the PersonalAI 2.0 framework, which introduces a planning-based dynamic graph traversal strategy. By integrating entity extraction, node matching, and LLM-driven query generation, the framework enables adaptive, iterative knowledge graph retrieval and reasoning. It combines graph traversal algorithms such as BeamSearch and WaterCircles, achieving an average 4% improvement across six QA benchmarks under LLM-as-a-Judge evaluation. The incorporation of planning yields an 18% performance gain, while graph traversal outperforms flat retrieval by 6%. Notably, PersonalAI 2.0 attains state-of-the-art results on MINE-1 with an 89% information retention rate.

0 citationsRead paper

Recent publications

Latest Papers

LLM-Guided Program Evolution for Targeted Black-Box Attacks on Perceptual Hash Algorithms

Jul 13, 2026

This work addresses the lack of provable robustness in perceptual hashing algorithms under adversarial perturbations by introducing GigaEvo/OpenEvolve, a large language model (LLM)-guided program evolution framework. For the first time, LLM-driven program synthesis is applied to black-box attacks on perceptual hashing, circumventing the need for internal algorithm access and effectively handling the discrete and non-differentiable nature of hash outputs. By optimizing a composite scoring metric, the method achieves high-efficiency attacks with minimal perturbation. Experimental results demonstrate that the approach significantly outperforms existing black-box attacks on pHash, PDQ, PhotoDNA, and NeuralHash, requiring fewer queries, achieving lower L2 distortion, and reducing the composite attack score by up to 41.2%, thereby exposing previously undisclosed security vulnerabilities in mainstream content moderation systems.

0 citationsRead paper

A Foundation Model for Multimodal Event Sequences in Financial Applications

Jul 10, 2026

This work addresses the limitations of traditional financial forecasting approaches, which rely on task-specific models and handcrafted features, struggle to integrate heterogeneous data (e.g., transactional and digital interaction logs), and exhibit poor reusability. The authors propose the first multimodal event-sequence foundation model tailored for financial applications. It unifies users’ multisource behaviors into temporal sequences and leverages a Transformer architecture pretrained via next-event prediction to learn general-purpose representations. These representations, combined with existing features, enable lightweight fine-tuning across diverse downstream tasks. The model achieves early fusion of multimodal events and cross-task representation sharing, effectively breaking down task silos. Deployed at a major bank in Eastern Europe, it significantly outperforms conventional models, enhancing key business metrics while reducing development costs.

0 citationsRead paper

A Stationary-Distribution Theory for Triplet-Based Plateau Search in Random Forest Ensemble-Size Selection

Jun 29, 2026

This work addresses the trade-off between computational cost and model stability in selecting the number of trees in random forests, noting that existing plateau-based tuning methods exhibit nondeterministic fluctuations even after apparent convergence. The study formulates ensemble size selection as a birth–death Markov chain on a geometric grid and derives its stationary distribution using local balance principles, revealing that plateau search is inherently a stochastic process. By introducing a symmetric correction update rule, the authors establish a theoretical framework for the stationary distribution of ternary plateau search, analyzed via a folded normal approximation centered at the optimum and a local Gaussian approximation. They prove that both the stationary center \(B_*\) and its standard deviation \(\sigma_{B,*}\) scale as \(O(\varepsilon^{-2})\), the variance as \(O(\varepsilon^{-4})\), and crucially, that the relative width of the distribution depends solely on the scaling factor and update rule, independent of the precision parameter \(\varepsilon\).

0 citationsRead paper

How Many Trees in a Random Forest? A Revisited Approach with Plateau Search and Optuna Integration

Jun 02, 2026

This work addresses the challenge of optimizing the number of trees in random forests, a hyperparameter whose performance typically improves monotonically with ensemble size, thereby confounding conventional hyperparameter optimization methods due to sensitivity to preset bounds or early-stopping noise. The authors propose a triplet plateau search algorithm that excludes this parameter from the Tree-structured Parzen Estimator (TPE) search space and instead employs a sliding window to monitor relative changes in out-of-bag (OOB) scores across three consecutive ensemble sizes. This approach adaptively identifies the minimal ensemble size satisfying a target performance criterion, eliminating the need for a predefined upper bound while offering both automation and interpretability. Theoretical analysis links the plateau criterion to the asymptotic performance gap and variance estimation. Empirical results demonstrate substantial reductions in tree count on most datasets, while automatically selecting larger ensembles on high-dimensional biological benchmarks such as Arcene and Dorothea, confirming the method’s adaptivity and effectiveness.

0 citationsRead paper

PersonalAI 2.0: Enhancing knowledge graph traversal/retrieval with planning mechanism for Personalized LLM Agents

May 13, 2026

Existing Graph Retrieval-Augmented Generation (GraphRAG) approaches for personalized large language model (LLM) agents suffer from static retrieval and a lack of planning mechanisms, leading to low factual accuracy and increased hallucination in question answering. To address these limitations, this work proposes the PersonalAI 2.0 framework, which introduces a planning-based dynamic graph traversal strategy. By integrating entity extraction, node matching, and LLM-driven query generation, the framework enables adaptive, iterative knowledge graph retrieval and reasoning. It combines graph traversal algorithms such as BeamSearch and WaterCircles, achieving an average 4% improvement across six QA benchmarks under LLM-as-a-Judge evaluation. The incorporation of planning yields an 18% performance gain, while graph traversal outperforms flat retrieval by 6%. Notably, PersonalAI 2.0 attains state-of-the-art results on MINE-1 with an 89% information retention rate.

0 citationsRead paper