arXiv subjects

Zhenyuan Huang

Publications and source records attributed to Zhenyuan Huang.

3 recordsLinked to original sources

arXiv2026

Thinking with Patterns: Breaking the Perceptual Bottleneck in Visual Planning via Pattern Induction

Planning from raw visual input remains a significant challenge for current Vision-Language Models (VLMs), when the complexity of input is beyond their one-step perception capability. Motivated by recent advances in Thinking with Images (TWI), a reasonable solution is to decompose the perception process into simpler steps by iteratively acquiring and incorporating local visual evidence. However, even though current VLMs are well-trained in general TWI ability, their perceptual bottleneck in the planning domain remains. To tackle this challenge, we formulate TWI as a tool to gradually build and reflect an accurate internal world model. We find that the resulting training-free planning strategy enables VLMs to solve tasks that are far beyond their initial capabilities, at the cost that too many TWI operations would significantly increase the computational overhead. To further improve efficiency, we propose Pattern Inference, a novel TWI strategy enabling VLMs to actively recognize known visual patterns in the new tasks and directly infer local world model structures. To obtain these patterns, we propose Pattern Induction, an online inductive learning strategy treating visual patterns as composite and reusable experts, which are autonomously discovered and optimized from experience. Experimental evaluations in FrozenLake, Crafter and CubeBench domains show that our approaches achieve a desirable balance between accuracy and efficiency.

cs.CV↗

arXiv2026

Worst-Case Sample Complexity Bounds for Distributed Inner Product Estimation with Local Randomized Measurements

We study distributed inner product estimation for $n$-qubit states using local randomized measurements, for which rigorous worst-case guarantees are less understood. We first reduce the minimax kernel optimization to Hamming-distance kernels. Within this class, unbiasedness fixes a unique kernel. For this kernel under local Clifford sampling, we prove a sharp fourth-moment bound using the single-qubit Clifford commutant. This yields worst-case sample complexity $\mathcal{O}(\sqrt{4.5^n})$, attained by identical pure product stabilizer states. For the same kernel under local Haar sampling, we prove a local twirling identity that compares its fourth moment with the Clifford fourth moment. This gives the same rigorous upper bound as in the Clifford case, but the comparison is lossy. This motivates the conjectured sharper Haar scaling $\mathcal{O}(\sqrt{3.6^n})$ attained by product states, and verify it for several important classes of states. We also show that independent single-qubit Pauli shadows have worst-case scaling $\mathcal{O}(\sqrt{7.5^n})$ for large $n$.

quant-ph↗

arXiv2026

FedSDWC: Federated Synergistic Dual-Representation Weak Causal Learning for OOD

Amid growing demands for data privacy and advances in computational infrastructure, federated learning (FL) has emerged as a prominent distributed learning paradigm. Nevertheless, differences in data distribution (such as covariate and semantic shifts) severely affect its reliability in real-world deployments. To address this issue, we propose FedSDWC, a causal inference method that integrates both invariant and variant features. FedSDWC infers causal semantic representations by modeling the weak causal influence between invariant and variant features, effectively overcoming the limitations of existing invariant learning methods in accurately capturing invariant features and directly constructing causal representations. This approach significantly enhances FL's ability to generalize and detect OOD data. Theoretically, we derive FedSDWC's generalization error bound under specific conditions and, for the first time, establish its relationship with client prior distributions. Moreover, extensive experiments conducted on multiple benchmark datasets validate the superior performance of FedSDWC in handling covariate and semantic shifts. For example, FedSDWC outperforms FedICON, the next best baseline, by an average of 3.04% on CIFAR-10 and 8.11% on CIFAR-100.

cs.LG↗