arXiv subjects

Tran Tien Dat

Publications and source records attributed to Tran Tien Dat.

3 recordsLinked to original sources

arXiv2026

Topologically protected chiral sensing using Synthetic Chiral Light

Chirality underlies molecular function in living matter, yet its optical detection remains challenging because conventional chiroptical spectroscopies rely on weak corrections to the dominant electric-dipole light-matter interaction, making desired optical signals weak and fragile. Topology offers a route to robustness, enabling observables whose defining properties survive disorder and imperfections. However, experimental realization of a practical topological observable for chiral spectroscopy has remained elusive. Here we realize chiral topological light and demonstrate such an observable. A tightly focused, phase-locked, counter-rotating two-colour field encodes chirality in the three-dimensional electric-field trajectory while its dominant topological charge resides in the longitudinal electric-field component where it remains hidden from direct far-field detection. An isotropic chiral medium acts as a topological transducer, converting the latent topology of the driving field into a propagating nonlinear response whose topological charge becomes directly observable in the far field while remaining strongly suppressed in achiral media. Using randomly oriented chiral single-crystal powders, we experimentally detect the enantio-sensitive response through the topological charge of the emitted field. Directly accessible in the far field and robust against experimental imperfections, this observable allows us to track and control chiral signal on attosecond timescales through the relative phase of the driving two-colour fields. Our results establish topology as a practical resource for ultrafast chiral optical spectroscopy.

physics.chem-ph↗

arXiv2026

HanoiWorld : A Joint Embedding Predictive Architecture BasedWorld Model for Autonomous Vehicle Controller

Current attempts of Reinforcement Learning for Autonomous Controller are data-demanding while the results are under-performed, unstable, and unable to grasp and anchor on the concept of safety, and over-concentrating on noise features due to the nature of pixel reconstruction. While current Self-Supervised Learningapproachs that learning on high-dimensional representations by leveraging the JointEmbedding Predictive Architecture (JEPA) are interesting and an effective alternative, as the idea mimics the natural ability of the human brain in acquiring new skill usingimagination and minimal samples of observations. This study introduces Hanoi-World, a JEPA-based world model that using recurrent neural network (RNN) formaking longterm horizontal planning with effective inference time. Experimentsconducted on the Highway-Env package with difference enviroment showcase the effective capability of making a driving plan while safety-awareness, with considerablecollision rate in comparison with SOTA baselines

cs.RO↗

arXiv2025

Mispronunciation Detection and Diagnosis Without Model Training: A Retrieval-Based Approach

Mispronunciation Detection and Diagnosis (MDD) is crucial for language learning and speech therapy. Unlike conventional methods that require scoring models or training phoneme-level models, we propose a novel training-free framework that leverages retrieval techniques with a pretrained Automatic Speech Recognition model. Our method avoids phoneme-specific modeling or additional task-specific training, while still achieving accurate detection and diagnosis of pronunciation errors. Experiments on the L2-ARCTIC dataset show that our method achieves a superior F1 score of 69.60% while avoiding the complexity of model training.

cs.CL↗