Searcharxiv⌕ Search

arXiv subjects

Jaemin Seo

Publications and source records attributed to Jaemin Seo.

14 recordsLinked to original sources
arXiv2026

Kernel-SDF: An Open-Source Library for Real-Time Signed Distance Function Estimation using Kernel Regression

Accurate and efficient scene representation is crucial for robotic tasks such as motion planning, manipulation, and navigation. Signed distance functions (SDFs) have emerged as a powerful representation for encoding distance to obstacle boundaries, enabling efficient collision-checking and trajectory optimization. However, existing methods are limited for large-scale uncertainty-aware SDF estimation from streaming sensor data: voxel-based approaches have fixed resolution and lack uncertainty quantification, neural network methods require significant training time, and Gaussian process (GP) methods struggle with scalability, sign estimation, and uncertainty calibration. In this letter, we develop an open-source library, Kernel-SDF, using kernel regression to learn SDF with calibrated uncertainty in real-time. It combines a front-end learning a continuous occupancy field via kernel regression with a back-end that estimates accurate SDF via GP regression using samples from the front-end surface boundaries. Kernel-SDF provides accurate SDF, gradient, uncertainty, and mesh construction in real-time. Evaluations show it achieves superior accuracy over existing methods while maintaining real-time performance, making it suitable for robotics tasks requiring reliable uncertainty-aware geometry.

cs.RO↗
arXiv2026

Deep Learning-based Surrogate Modelling of the LOD Method for Multiscale Problems

Multiscale problems are notoriously difficult to tackle using traditional numerical methods, as accurately resolving fine-scale features often requires prohibitively fine discretizations. This challenge is particularly pronounced in applications such as materials science, fluid dynamics, climate systems, chemical processes, and complex networks. Recent neural operator models provide a promising data-driven alternative, but frequently struggle to achieve sufficient accuracy in the presence of strongly heterogeneous or oscillatory coefficients. In this work, we focus on the solution of elliptic PDEs with rough and high-contrast inputs. The Localized Orthogonal Decomposition (LOD) method is a well-established numerical approach for such problems, but it comes, however, at a substantial computational cost. We investigate the performance of popular neural operator architectures on these challenging multiscale problems and identify key limitations in their ability to resolve fine-scale structure. To overcome these challenges, we introduce LOD-MSNO (LOD-Multiscale Neural Operator), a hybrid approach that leverages the LOD method as a strong multiscale prior by building on its representation of the solution as a linear combination of problem-adapted basis functions, while addressing its main computational bottlenecks through data-driven operator learning. We further provide theoretical error estimates for the proposed coefficient-learning framework. Lastly, we demonstrate the potential of our proposed method to outperform current neural operator baselines in terms of accuracy for challenging multiscale inputs, while mainly retaining the computational efficiency of neural operator models.

math.NA↗
arXiv2026

TokaGrad: End-to-end differentiable tokamak simulator for L-to-H full scenario optimization

As fusion energy moves from theoretical feasibility toward commercialization, design of new reactor concepts, autonomous tokamak control, and high-performance scenario optimization are becoming increasingly important. Traditionally, such optimization tasks have relied on costly trial-and-error or brute-force parameter searches, based on black-box experiments or simulations. Recently, advances in differentiable programming are changing the paradigm of numerical simulation. Unlike conventional simulations, which are typically executed as locally connected step-by-step procedures, differentiable simulation represents the entire simulation pipeline as a connected computational graph. In such a framework, machine parameters, actuator waveforms, and plasma responses are linked through differentiable operations, allowing Jacobians to propagate across the full simulation. This enables direct gradient-based control and optimization using the internal sensitivities of the simulator, rather than treating the simulator as a black box. Here, we present TokaGrad, an end-to-end differentiable tokamak transport simulator for full-scenario modeling, including ramp-up, L-mode operation, and H-mode access. TokaGrad self-consistently integrates differentiable models for plasma equilibrium, transport, heating, L-H transition, and pedestal formation. To our knowledge, this is the first differentiable tokamak simulator capable of self-consistently modeling dynamic full-discharge scenarios where actuators and plasma evolve together with equilibrium, pedestal, and confinement-regime transitions. We demonstrate that, when coupled to gradient-based optimizers, TokaGrad enables reactor-design optimization, actuator control, and full-scenario waveform optimization. This framework provides a pathway toward automated, differentiable optimization of burning-plasma scenarios and reactor concepts.

math-ph↗
arXiv2026

Unbiased and Second-Order-Free Training for High-Dimensional PDEs

Deep learning methods based on backward stochastic differential equations (BSDEs) have emerged as competitive alternatives to physics-informed neural networks (PINNs) for solving high-dimensional partial differential equations (PDEs). By leveraging probabilistic representations, BSDE approaches can avoid the curse of dimensionality and often admit second-order-free training objectives that do not require explicit Hessian evaluations. It has recently been established that the commonly used Euler-Maruyama (EM) time discretization induces an intrinsic bias in BSDE training losses. While high-order schemes such as Heun can fully eliminate this bias, such schemes re-introduce second-order spatial derivatives and incur substantial computational overhead. In this work, we provide a principled analysis of EM-induced loss bias and propose an unbiased, second-order-free training framework that preserves the computational advantages of BSDE methods. Our code is available at https://github.com/seojaemin22/Un-EM-BSDE.

cs.LG↗
arXiv2026

Single-shot quantum neural networks with amplitude estimation

Quantum neural networks (QNNs) suffer from a fundamental sampling bottleneck since quantum measurements are probabilistic, requiring many circuit executions to estimate outputs with sufficient accuracy. Conventional Monte-Carlo (MC) inference exhibits an $\mathcal{O}(1/\sqrt{N})$ sampling error, rendering QNN inference and training costly on near-term quantum hardware, especially where each shot requires expensive qubit generation. This work introduces a "single-shot" QNN framework by integrating quantum amplitude estimation (AE) into the readout stage. By embedding a trained QNN as a state-preparation oracle within AE, outputs are estimated through coherent interference rather than repeated sampling. We demonstrate that AE-based QNN inference achieves an $\mathcal{O}(1/N)$ error even with a single shot. We further analyze noise robustness and training feasibility, showing that AE can be a powerful primitive for overcoming the sampling overhead of QNNs. This highlights that when the model itself is quantum, quantum algorithms can enhance the computation efficiency.

quant-ph↗
arXiv2026

Efficient quantum machine learning with inverse-probability algebraic corrections

Quantum neural networks (QNNs) provide expressive probabilistic models by leveraging quantum superposition and entanglement, yet their practical training remains challenging due to highly oscillatory loss landscapes and noise inherent to near-term quantum devices. Existing training approaches largely rely on gradient-based procedural optimization, which often suffers from slow convergence, sensitivity to hyperparameters, and instability near sharp minima. In this work, we propose an alternative inverse-probability algebraic learning framework for QNNs. Instead of updating parameters through incremental gradient descent, our method treats learning as a local inverse problem in probability space, directly mapping discrepancies between predicted and target Born-rule probabilities to parameter corrections via a pseudo-inverse of the Jacobian. This algebraic update is covariant, does not require learning-rate tuning, and enables rapid movement toward the vicinity of a loss minimum in a single step. We systematically compare the proposed method with gradient descent and Adam optimization in both regression and classification tasks using a teacher-student QNN benchmark. Our results show that algebraic learning converges significantly faster, escapes loss plateaus, and achieves lower final errors. Under finite-shot sampling, the method exhibits near-optimal error scaling, while remaining robust against intrinsic hardware noise such as dephasing. These findings suggest that inverse-probability algebraic learning offers a principled and practical alternative to procedural optimization for QNN training, particularly in resource-constrained near-term quantum devices.

quant-ph↗
arXiv2026

Communication-Free Collective Navigation for a Swarm of UAVs via LiDAR-Based Deep Reinforcement Learning

This paper presents a deep reinforcement learning (DRL) based controller for collective navigation of unmanned aerial vehicle (UAV) swarms in communication-denied environments, enabling robust operation in complex, obstacle-rich environments. Inspired by biological swarms where informed individuals guide groups without explicit communication, we employ an implicit leader-follower framework. In this paradigm, only the leader possesses goal information, while follower UAVs learn robust policies using only onboard LiDAR sensing, without requiring any inter-agent communication or leader identification. Our system utilizes LiDAR point clustering and an extended Kalman filter for stable neighbor tracking, providing reliable perception independent of external positioning systems. The core of our approach is a DRL controller, trained in GPU-accelerated Nvidia Isaac Sim, that enables followers to learn complex emergent behaviors - balancing flocking and obstacle avoidance - using only local perception. This allows the swarm to implicitly follow the leader while robustly addressing perceptual challenges such as occlusion and limited field-of-view. The robustness and sim-to-real transfer of our approach are confirmed through extensive simulations and challenging real-world experiments with a swarm of five UAVs, which successfully demonstrated collective navigation across diverse indoor and outdoor environments without any communication or external localization.

cs.RO↗
arXiv2025

Quantum mixture-density network for multimodal probabilistic prediction

Multimodal probability distributions are common in both quantum and classical systems, yet modeling them remains challenging when the number of modes is large or unknown. Classical methods such as mixture-density networks (MDNs) scale poorly, requiring parameter counts that grow quadratically with the number of modes. We introduce a Quantum Mixture-Density Network (Q-MDN) that employs parameterized quantum circuits to efficiently model multimodal distributions. By representing an exponential number of modes with a compact set of qubits and parameters, Q-MDN predicts Gaussian mixture components with high resolution. We evaluate Q-MDN on two benchmark tasks: the quantum double-slit experiment and chaotic logistic bifurcation. In both cases, Q-MDN outperforms classical MDNs in mode separability and prediction sharpness under equal parameter budgets. Our results demonstrate an efficiency in probabilistic regression and highlight the potential of quantum machine learning in capturing complex stochastic behavior.

quant-ph↗
arXiv2025

PINT: Physics-Informed Neural Time Series Models with Applications to Long-term Inference on WeatherBench 2m-Temperature Data

This paper introduces PINT (Physics-Informed Neural Time Series Models), a framework that integrates physical constraints into neural time series models to improve their ability to capture complex dynamics. We apply PINT to the ERA5 WeatherBench dataset, focusing on long-term forecasting of 2m-temperature data. PINT incorporates the Simple Harmonic Oscillator Equation as a physics-informed prior, embedding its periodic dynamics into RNN, LSTM, and GRU architectures. This equation's analytical solutions (sine and cosine functions) facilitate rigorous evaluation of the benefits of incorporating physics-informed constraints. By benchmarking against a linear regression baseline derived from its exact solutions, we quantify the impact of embedding physical principles in data-driven models. Unlike traditional time series models that rely on future observations, PINT is designed for practical forecasting. Using only the first 90 days of observed data, it iteratively predicts the next two years, addressing challenges posed by limited real-time updates. Experiments on the WeatherBench dataset demonstrate PINT's ability to generalize, capture periodic trends, and align with physical principles. This study highlights the potential of physics-informed neural models in bridging machine learning and interpretable climate applications. Our models and datasets are publicly available on GitHub: https://github.com/KV-Park.

cs.LG↗
arXiv2025

Multivariate unbounded quantum regression via log-ratio probabilities mitigating barren plateaus

Quantum neural networks (QNNs) have shown remarkable potential due to their capability of representing complex functions within exponentially large Hilbert spaces. However, their application to multivariate regression tasks has been limited, primarily due to inherent constraints of traditional approaches that rely on Pauli expectation values. In this work, we introduce a novel and simple post-processing method utilizing log-ratio probabilities (LRPs) of quantum states, enabling efficient and unbounded multivariate regression within existing QNN architectures. Our approach exponentially expands the number of regression outputs relative to qubit count, thus significantly improving parameter and qubit efficiency. Additionally, by enhancing parameter dependencies in the cost function and leveraging gradient pumping effects from the log-ratio transformation, our method mitigates the well-known barren plateau phenomenon, thereby stabilizing training. We further demonstrate that this approach facilitates robust uncertainty quantification, capturing both epistemic and aleatoric uncertainties simultaneously. Our findings underscore the practical potential of LRP-QNNs for complex multi-output regression tasks, particularly within current resource-constrained quantum hardware environments.

quant-ph↗
arXiv2025

Kalman Filter-Based Distributed Gaussian Process for Unknown Scalar Field Estimation in Wireless Sensor Networks

In this letter, we propose an online scalar field estimation algorithm of unknown environments using a distributed Gaussian process (DGP) framework in wireless sensor networks (WSNs). While the kernel-based Gaussian process (GP) has been widely employed for estimating unknown scalar fields, its centralized nature is not well-suited for handling a large amount of data from WSNs. To overcome the limitations of the kernel-based GP, recent advancements in GP research focus on approximating kernel functions as products of E-dimensional nonlinear basis functions, which can handle large WSNs more efficiently in a distributed manner. However, this approach requires a large number of basis functions for accurate approximation, leading to increased computational and communication complexities. To address these complexity issues, the paper proposes a distributed GP framework by incorporating a Kalman filter scheme (termed as K-DGP), which scales linearly with the number of nonlinear basis functions. Moreover, we propose a new consensus protocol designed to handle the unique data transmission requirement residing in the proposed K-DGP framework. This protocol preserves the inherent elements in the form of a certain column in the nonlinear function matrix of the communicated message; it enables wireless sensors to cooperatively estimate the environment and reach the global consensus through distributed learning with faster convergence than the widely-used average consensus protocol. Simulation results demonstrate rapid consensus convergence and outstanding estimation accuracy achieved by the proposed K-DGP algorithm. The scalability and efficiency of the proposed approach are further demonstrated by online dynamic environment estimation using WSNs.

cs.MA↗
arXiv2024

Multimodal Super-Resolution: Discovering hidden physics and its application to fusion plasmas

A non-linear system governed by multi-spatial and multi-temporal physics scales cannot be fully understood with a single diagnostic, as each provides only a partial view, leading to information loss. Combining multiple diagnostics may also result in incomplete projections of the system's physics. By identifying hidden inter-correlations between diagnostics, we can leverage mutual support to fill in these gaps, but uncovering such correlations analytically is too complex. We introduce a machine learning methodology to address this issue. Unlike traditional methods, our multimodal approach does not rely on the target diagnostic's direct measurements to generate its super-resolution version. Instead, it uses other diagnostics to produce super-resolution data, capturing detailed structural evolution and responses to perturbations previously unobservable. This not only enhances the resolution of a diagnostic for deeper insights but also reconstructs the target diagnostic, providing a valuable tool to mitigate diagnostic failure. This methodology addresses a key challenge in fusion plasmas: the Edge Localized Mode (ELM), a plasma instability that can cause significant erosion of plasma-facing materials. A method to stabilize ELM is using resonant magnetic perturbation (RMP) to trigger magnetic islands. However, limited spatial and temporal resolution restricts analysis of these islands due to their small size, rapid dynamics, and complex plasma interactions. With super-resolution diagnostics, we can experimentally verify theoretical models of magnetic islands for the first time, providing insights into their role in ELM stabilization. This advancement supports the development of effective ELM suppression strategies for future fusion reactors like ITER and has broader applications, potentially revolutionizing diagnostics in fields such as astronomy, astrophysics, and medical imaging.

physics.plasm-ph↗
arXiv2024

Design Optimization of Nuclear Fusion Reactor through Deep Reinforcement Learning

This research explores the application of Deep Reinforcement Learning (DRL) to optimize the design of a nuclear fusion reactor. DRL can efficiently address the challenging issues attributed to multiple physics and engineering constraints for steady-state operation. The fusion reactor design computation and the optimization code applicable to parallelization with DRL are developed. The proposed framework enables finding the optimal reactor design that satisfies the operational requirements while reducing building costs. Multi-objective design optimization for a fusion reactor is now simplified by DRL, indicating the high potential of the proposed framework for advancing the efficient and sustainable design of future reactors.

physics.plasm-ph↗
arXiv2023

Enhancing Disruption Prediction through Bayesian Neural Network in KSTAR

Disruption in tokamak plasmas, stemming from various instabilities, poses a critical challenge, resulting in detrimental effects on the associated devices. Consequently, the proactive prediction of disruptions to maintain stability emerges as a paramount concern for future fusion reactors. While data-driven methodologies have exhibited notable success in disruption prediction, conventional neural networks within a frequentist approach cannot adequately quantify the uncertainty associated with their predictions, leading to overconfidence. To address this limit, we utilize Bayesian deep probabilistic learning to encompass uncertainty and mitigate false alarms, thereby enhancing the precision of disruption prediction. Leveraging 0D plasma parameters from EFIT and diagnostic data, a Temporal Convolutional Network adept at handling multi-time scale data was utilized. The proposed framework demonstrates proficiency in predicting disruptions, substantiating its effectiveness through successful applications to KSTAR experimental data.

physics.plasm-ph↗