1

We propose AgentCBO, a causal belief routing framework that brings LLM causal priors into causal Bayesian optimization through guarded proposal and commit mechanisms, performing comparably to known-graph CBO across ten benchmarks.

We show that reasoning confidence is not policy-invariant and propose CPCC, a policy-conditioned calibrator of prefix success that attains the lowest Brier score across six continuation policies and three reasoning benchmarks.

We propose theoretically sound stopping criteria for GP-UCB based on tighter instantaneous regret bounds, ensuring an $\epsilon$-optimal solution with high probability upon termination.

We introduce CCGBO, a Bayesian Optimization framework that uses counterfactual credit to quantify historical observations’ contribution to discovering the optimum, accelerating convergence.

We propose THCM-CAL, a temporal-hierarchical causal modeling framework with conformal calibration for clinical risk prediction from structured and unstructured electronic health records.

We propose SynGraph, a dynamic Graph-LLM synthesis framework that mitigates data sparsity in streaming sentiment analysis using Continuous-Time Dynamic Graphs and LLM-augmented representations.

We propose a deeply supervised semantic segmentation model built on a GAN framework, improving road crack segmentation over existing methods such as SEGAN.

We propose a face synthesis approach that controls facial attributes by matching latent vectors to attributes and disentangling the attribute feature axes in latent space with Gram-Schmidt orthogonalization.

We propose a semi-supervised wafer map defect classification method that uses latent representations from a pre-trained VAE to guide teacher-student training, outperforming state-of-the-art models on the WM-811K dataset.

We formulate Bayesian optimization with path-dependent switching costs as a constrained MDP and propose distUCB, a distance-adjusted multi-step look-ahead acquisition function with a theoretical regret analysis.