Kyunghyun Park

Daewon Choi, Kyunghyun Park, Woomin Song et al.

Large language model (LLM)-based agents solve complex tasks by leveraging multi-step reasoning with iterative tool calls and environment interactions, which incur idle time while waiting for observations. Despite the prevalence of idle time in most agentic scenarios, existing works treat it as an unavoidable overhead or propose restricted solutions that overlook varying computational budgets across different tool calls and future observation uncertainty, thereby leading to suboptimal utilization of idle time. In this paper, we introduce IdleSpec, a scalable and generic inference approach that leverages idle-time computation to improve agent performance while minimizing latency overhead. Specifically, IdleSpec iteratively generates plan candidates during idle periods and, once observations become available, aggregates them to guide the next reasoning step. For effective plan generation under observation uncertainty, IdleSpec samples between complementary drafting strategies (i.e., progressive and recovery) from a learned distribution that is updated via posterior feedback. Our experiments demonstrate that IdleSpec significantly improves agent performance in various agentic scenarios by effectively utilizing idle time. In particular, on the GAIA and FRAMES, IdleSpec achieves 55.6% average accuracy with Gemini-2.5-Flash, surpassing the vanilla baseline without idle-time usage by 5.1%. Furthermore, for MLE-Bench, which involves substantial delay from code executions, IdleSpec achieves performance gains of up to 9.1% on the Any Medal rate, highlighting its generalizability to long-horizon tasks.

Daniel Bartl, Ariel Neufeld, Kyunghyun Park

We examine nonlinear Kolmogorov partial differential equations (PDEs). Here the nonlinear part of the PDE comes from its Hamiltonian where one maximizes over all possible drift and diffusion coefficients which fall within a $\varepsilon$-neighborhood of pre-specified baseline coefficients. Our goal is to quantify and compute how sensitive those PDEs are to such a small nonlinearity, and then use the results to develop an efficient numerical method for their approximation. We show that as $\varepsilon\downarrow 0$, the nonlinear Kolmogorov PDE equals the linear Kolmogorov PDE defined with respect to the corresponding baseline coefficients plus $\varepsilon$ times a correction term which can be also characterized by the solution of another linear Kolmogorov PDE involving the baseline coefficients. As these linear Kolmogorov PDEs can be efficiently solved in high-dimensions by exploiting their Feynman-Kac representation, our derived sensitivity analysis then provides a Monte Carlo based numerical method which can efficiently solve these nonlinear Kolmogorov equations. We establish an error and complexity analysis for our numerical method. Moreover, we provide numerical examples in up to 100 dimensions to empirically demonstrate the applicability of our numerical method.

Joonhun Lee, Myeongho Jeon, Myungjoo Kang et al.

We propose Feature-aligned N-BEATS as a domain-generalized time series forecasting model. It is a nontrivial extension of N-BEATS with doubly residual stacking principle (Oreshkin et al. [45]) into a representation learning framework. In particular, it revolves around marginal feature probability measures induced by the intricate composition of residual and feature extracting operators of N-BEATS in each stack and aligns them stack-wise via an approximate of an optimal transport distance referred to as the Sinkhorn divergence. The training loss consists of an empirical risk minimization from multiple source domains, i.e., forecasting loss, and an alignment loss calculated with the Sinkhorn divergence, which allows the model to learn invariant features stack-wise across multiple source data sequences while retaining N-BEATS's interpretable design and forecasting power. Comprehensive experimental evaluations with ablation studies are provided and the corresponding results demonstrate the proposed model's forecasting and generalization capabilities.