arXiv subjects

Junghyun Kim

Publications and source records attributed to Junghyun Kim.

At least 19 recordsLinked to original sources

arXiv2026

Multi-Backbone Self-Supervised Ensembles for Audio Deepfake Detection and a Cross-Track Analysis of Generation-Detection Asymmetry

This paper describes the participation of team "Go-To-Germany" in the ImageCLEF 2026 Audio Deepfake Detection and Generation task. Our detection system, built on a four-backbone self-supervised learning (SSL) ensemble combining WavLM-Large, Wav2Vec2-XLS-R-300M, ECAPA-TDNN, and x-vector representations, achieved a final score of 0.9522 on the official ImageCLEF 2026 evaluation, with perfect accuracy (1.0000) on participant-generated deepfakes and 0.8875 on the held-out organizer ground-truth real data. For the Generation sub-task, our official team submission, an F5-TTS v1 baseline processed with a uniform reverberation pass and submitted as a deliberate anti-forensic probe, ranked first with a final score of 0.4304 (word error rate (WER) 4.99%, character error rate (CER) 2.07%); details of our four-model program (GLM-TTS, F5-TTS, XTTS v2, CosyVoice3), from which the official entry was drawn, appear in the paper. We present a cross-track analysis revealing a pronounced asymmetry: our detection system identifies 100% of participant-generated deepfakes, while our official generation entry, despite ranking first in the Audio Generation sub-task and evading 61.4% and 56.2% of participant and organizer detectors, attains a Final Score of 0.4304 against 0.9522 on the Detection side. We further report falsification-based ablation experiments (LOSO 56-speaker cross-validation, three-region backbone geometry, bootstrap confidence intervals, and PCA analysis) that motivate our architectural-insurance hypothesis for multi-backbone SSL ensembling. We complement these results with five cross-track insights and five pre-registered falsification experiments connecting generation-side evasion to detection-side design decisions, and we openly report an 11.25% false-positive gap on held-out organizer real recordings as the principal open challenge for deployment.

cs.SD↗

arXiv2026

Unconventional and Fragile Magnetic Exciton in a van der Waals Quantum Magnet

The recently discovered magnetic exciton in the van der Waals (vdW) antiferromagnet NiPS3 exemplifies these phenomena, exhibiting several distinctive characteristics. Despite extensive investigation, much of its physics remains unresolved, with key questions about why the NiPS3 magnetic exciton is so sharp and optically bright despite the nominally spin-forbidden transition, posing significant challenges to a proper understanding and practical manipulation of the exciton. An urgent question is to what extent it is due to chemical disorder, magnetic weakening, lattice modification, or intrinsic instability of the bright exciton itself: answers to which will put stringent constraints on possible theoretical models. Here we address these questions using hydrostatic pressure as a clean, continuous, reversible, and in-situ tuning parameter. We find that the sharp photoluminescence peak is drastically suppressed by as little as 0.4 GPa and completely quenched by 1.5 GPa, with demonstrating its reversibility. Crucially, this bright-to-dark conversion occurs without magnetic, crystallographic, or electronic reconstruction despite an increase in the Neel temperature, as established by Raman, X-ray absorption, nuclear magnetic resonance spectroscopy, and first-principles many-body calculations. Our results demonstrate that the optical brightness of the magnetic exciton is independent of chemical disorder, lattice expansion, and weakening of magnetic order, indicating that a higher-order correlated mechanism governs the bright exciton. We further propose experimentally constrained microscopic scenarios involving exciton pairing, crystal-field-controlled spin-orbit mixing, and symmetry breaking, providing a framework for future tests of entangled magnetic exciton in correlated quantum magnets.

cond-mat.mtrl-sci↗

arXiv2026

Hardware-efficient erasure-error detection with an integer fluxonium

Erasure-error detection can improve the efficiency of quantum error correction by revealing the times and locations of their error events. In this work, we demonstrate erasure conversions and mid-circuit erasure detections in a single integer fluxonium, in which the states $\mathrm{|g\rangle, |f\rangle}$ encode the logical states and $\mathrm{|e\rangle}$ encodes the erasure state. The integer fluxonium suppresses direct $|\mathrm{f} \rangle \rightarrow |\mathrm{g} \rangle$ transitions and allows the dominant $|\mathrm{f} \rangle \rightarrow |\mathrm{e}\rangle$ transitions to be converted into detectable erasures. Furthermore, we identified a design space that nullifies the resonant-frequency shift between the two logical states, enabling ancilla-free mid-circuit erasure checks using the same resonator employed for final readout. By discarding the detected erasure events, we achieved an 8.4-fold increase in the $|\mathrm{f}\rangle$ state lifetime, a 1.38-fold increase in the Hahn-echo time, and a reduction of single-qubit gate error from 0.061(2)% to 0.030(5)%. Our results establish integer fluxonium as a hardware-efficient platform for erasure-error detection and conversion, while identifying the improvements required to realize an effective erasure qubit with high erasure bias.

quant-ph↗

arXiv2026

Adversarial Deepfake Generation and an Investigation of Purification-Based Adversarial Detection

This paper describes the participation of team "Go To Germany" in the ImageCLEF 2026 Deepfake Detection and Generation Task. For the image generation task, we employ FLUX.1-dev with PuLID for identity-preserving face synthesis, combined with a multi-model PGD adversarial attack targeting 12 detectors simultaneously (DiffJPEG-in-loop, MI/DI/EoT, adaptive weighting, two-stage warm-start). Our approach achieved 90% evasion against organizer detectors and 57.6% against participant detectors, with a final generation score of 0.4170. For the image detection task, we combine two complementary detectors - SigLIP+DINOv2 for AI-generated images and GenD-DINOv3 for face manipulations - in a max-probability ensemble, achieving 99.4% accuracy on baseline deepfakes but suffering from high false-positive rates on real images, resulting in a final detection score of 0.6986. Beyond the official submission, we conducted a self-initiated investigation of purification-based adversarial detection, comparing three families of detection signals across six detectors that share a CLIP ViT-L/14 backbone. We find that raw $|\Delta \text{logit}|$ under median-3 purification, applied through the EFFORT detector, separates adversarial inputs from clean inputs with AUROC 0.81-0.98 across four adversarial source types - a finding that refutes the simple backbone-preservation hypothesis and exposes a sharp JPEG-quality cliff at Q70 where the signal collapses.

cs.CV↗

arXiv2026

Microscopic evidence for a Zhang-Rice triplet state in the van der Waals antiferromagnet, NiPS$_3$

Quantum-entangled states underpin many emergent phenomena in quantum materials, yet their direct experimental identification remains a challenge. NiPS$_3$, a van der Waals antiferromagnet exhibiting a resolution-limited magnetic exciton in its ordered phase, has been proposed to host a many-body entangled Zhang-Rice triplet state. Here, using $^{33}$S nuclear magnetic resonance (NMR) on $^{33}$S-enriched NiPS$_3$ single crystals, we provide microscopic evidence for this charge-transfer state. The $^{33}$S and $^{31}$P Knight shifts as a function of temperature reveal a unified spin-triplet configuration arising from strong hybridization between a self-doped hole in the S $3p$ orbitals and a hole in Ni $3d$ orbitals. Furthermore, the $^{33}$S nuclear spin-lattice relaxation rate exhibits a power-law divergence as it approaches the N\'eel temperature $T_N=155$ K, indicating critical slowing down of collective charge fluctuations consistent with spin-nematic correlations. These results reveal a spin-charge-intertwined ground state and establish the microscopic foundation for the exceptional coherence of the magnetic exciton in NiPS$_3$.

cond-mat.str-el↗

arXiv2026

PHASOR: Phase-Anchored Universal Action Representations for Humanoid Embodiments

Learning a good action embedding space is fundamental to scalable robot policy learning, yet existing methods treat action latents as task-specific intermediates rather than first-class representations. The resulting latents are unstructured, embodiment-specific, and weakly tied to motion semantics, limiting interpretability, controllability, and transferability across robots. We position the action embedding space itself as a first-class design target, with downstream policy quality emerging from representation quality. Exploiting motion's intrinsic periodicity, we factorize it into a phase manifold that captures cyclic structure via FFT-parametric coefficients, together with a pose branch that conditions the manifold on non-periodic configuration detail. Combined with motion-semantic distillation, this factorized structure yields a cross-embodiment motion manifold that is interpretable and embodiment-agnostic by design. Anchoring multiple humanoid robots to a shared human-pretrained manifold then produces a unified action embedding space across diverse platforms, achieving strong cross-embodiment retrieval and consistent gains on downstream robot tasks.

cs.RO↗

arXiv2026

Characterization and Comparison of Energy Relaxation in Fluxonium Qubits

Fluxonium superconducting qubits have demonstrated long coherence times and high single- and two-qubit gate fidelities, making them a favorable building block for superconducting quantum processors. We investigate the dominant limitations to fluxonium qubit energy relaxation time $T_1$ using a set of eight planar, aluminum-on-silicon qubits. We find that a circuit-based model for capacitive dielectric loss best captures the frequency dependence of $T_1$, which we analyze within both a two-level and a six-level energy relaxation model. We convert the measured $T_1$ into an effective capacitive quality factor $Q_\mathrm{C}^{\mathrm{eff}}$ to compare qubits on equal footing, accounting for independently estimated contributions from $1/f$ flux noise and radiative loss to the control and readout circuitry. We apply this methodology to compare qubits from two fabrication processes: a baseline process and one that applies a fluorine-based wet treatment prior to Josephson junction deposition. We resolve a small improvement of (13.8 $\pm$ 8.4$)\%$ in the process mean $Q_\mathrm{C}^{\mathrm{eff}}$, indicating that the fluorine treatment may have reduced loss from the metal-substrate interface, but did not address the primary source of loss in these fluxonium qubits.

quant-ph↗

arXiv2025

ZZ-Free Two-Transmon CZ Gate Mediated by a Fluxonium Coupler

Eliminating residual ZZ interactions in a two-qubit system is essential for reducing coherent errors during quantum operations. In a superconducting circuit platform, coupling two transmon qubits via a transmon coupler has been shown to effectively suppress residual ZZ interactions. However, in such systems, perfect cancellation usually requires the qubit-qubit detuning to be smaller than the individual qubit anharmonicities, which exacerbates frequency crowding and microwave crosstalk. To address this limitation, we introduce TFT (Transmon-Fluxonium-Transmon) architecture, wherein two transmon qubits are coupled via a fluxonium qubit. The coupling mediated by the fluxonium eliminates residual ZZ interactions even for transmons detuned larger than their anharmonicities. We experimentally identified zero-ZZ interaction points at qubit-qubit detunings of 409 MHz and 616 MHz from two distinct TFT devices. We then implemented an adiabatic, coupler-flux-biased controlled-Z gate on both devices, achieving CZ gate fidelities of 99.64(6)% and 99.68(8)%.

quant-ph↗

arXiv2025

EgoExo-Con: Exploring View-Invariant Video Temporal Understanding

Do Video-LLMs have consistent temporal understanding when videos capture the same event from different viewpoints? To study this question, we introduce EgoExo-Con(sistency), a benchmark of synchronized egocentric and exocentric video pairs with human-refined queries that ensure all concepts are visible in both viewpoints. EgoExo-Con emphasizes two temporal understanding tasks: Temporal Verification and Temporal Grounding. It evaluates not only correctness but consistency across viewpoints. Our analysis reveals two critical limitations of existing Video-LLMs: (1) models often fail to maintain consistency, with results far worse than their single-view performances. (2) When naively finetuned with synchronized videos of both viewpoints, the models show improved consistency but often underperform those trained on a single view. For improvements, we propose View-GRPO, a novel reinforcement learning framework that effectively strengthens view-specific temporal reasoning while encouraging consistent comprehension across viewpoints. Our method demonstrates its superior temporal understanding capabilities, especially for improving cross-view consistency. All resources have been made available at https://minjoong507.github.io/projects/EgoExo-Con/

cs.CV↗

arXiv2025

Coherent terahertz control of metastable magnetization in FePS3

The crystal lattice governs the emergent electronic, magnetic, and optical properties of quantum materials, making structural tuning through strain, pressure, or chemical substitution a key approach for discovering and controlling novel quantum phases. Beyond static modifications, driving specific lattice modes with ultrafast stimuli offers a dynamic route for tailoring material properties out of equilibrium. However, achieving dynamic coherent control of the nonequilibrium phases via resonant excitation of lattice coherences remains largely unexplored. Such manipulation enables non-volatile, on demand amplification and suppression of order parameters on femtosecond timescales, necessary for next generation optoelectronic ultrafast computation. In this study, we demonstrate coherent phononic control of a newly discovered, light-induced metastable magnetization in the van der Waals antiferromagnet FePS3. By using a sequence of terahertz (THz) pulses, we modulate the magnetization amplitude at the frequencies of phonon coherences, whose infrared-active nature and symmetries are further revealed by polarization- and field-strength-dependent measurements. Furthermore, our two-dimensional THz spectroscopy, in tandem with first-principles numerical simulations, shows that these phonons nonlinearly displace a Raman active phonon, which induces the metastable net magnetization. These findings not only clarify the microscopic mechanism underlying the metastable state in FePS3 but also establish vibrational coherences in solids as a powerful tool for ultrafast quantum phase control, enabling manipulation of material functionalities far from equilibrium.

cond-mat.mtrl-sci↗

arXiv2025

Emergent Harmonics in Josephson Tunnel Junctions Due to Series Inductance

Josephson tunnel junctions are essential elements of superconducting quantum circuits. The operability of these circuits presumes a $2\pi$-periodic sinusoidal potential of a tunnel junction, but higher-order corrections to this Josephson potential, often referred to as "harmonics," cause deviations from the expected circuit behavior. Two potential sources for these harmonics are the intrinsic current-phase relationship of the Josephson junction and the inductance of the metallic traces connecting the junction to other circuit elements. Here, we introduce a method to distinguish the origin of the observed harmonics using nearly-symmetric superconducting quantum interference devices (SQUIDs). Spectroscopic measurements of level transitions in multiple devices reveal features that cannot be explained by a standard cosine potential, but are accurately reproduced when accounting for a second-harmonic contribution to the model. The observed scaling of the second harmonic with Josephson-junction size indicates that it is due almost entirely to the trace inductance. These results inform the design of next-generation superconducting circuits for quantum information processing and the investigation of the supercurrent diode effect.

quant-ph↗

arXiv2025

Terahertz field-induced metastable magnetization near criticality in FePS3

Controlling the functional properties of quantum materials with light has emerged as a frontier of condensed-matter physics, leading to the discovery of various light-induced phases of matter, such as superconductivity, ferroelectricity, magnetism and charge density waves. However, in most cases, the photoinduced phases return to equilibrium on ultrafast timescales after the light is turned off, limiting their practical applications. Here we use intense terahertz pulses to induce a metastable magnetization with a remarkably long lifetime of more than 2.5 milliseconds in the van der Waals antiferromagnet FePS3. The metastable state becomes increasingly robust as the temperature approaches the antiferromagnetic transition point, suggesting that critical order parameter fluctuations play an important part in facilitating the extended lifetime. By combining first-principles calculations with classical Monte Carlo and spin dynamics simulations, we find that the displacement of a specific phonon mode modulates the exchange couplings in a manner that favours a ground state with finite magnetization near the N\'eel temperature. This analysis also clarifies how the critical fluctuations of the dominant antiferromagnetic order can amplify both the magnitude and the lifetime of the new magnetic state. Our discovery demonstrates the efficient manipulation of the magnetic ground state in layered magnets through non-thermal pathways using terahertz light and establishes regions near critical points with enhanced order parameter fluctuations as promising areas to search for metastable hidden quantum states.

cond-mat.mtrl-sci↗

arXiv2025

Socratic Planner: Self-QA-Based Zero-Shot Planning for Embodied Instruction Following

Embodied Instruction Following (EIF) is the task of executing natural language instructions by navigating and interacting with objects in interactive environments. A key challenge in EIF is compositional task planning, typically addressed through supervised learning or few-shot in-context learning with labeled data. To this end, we introduce the Socratic Planner, a self-QA-based zero-shot planning method that infers an appropriate plan without any further training. The Socratic Planner first facilitates self-questioning and answering by the Large Language Model (LLM), which in turn helps generate a sequence of subgoals. While executing the subgoals, an embodied agent may encounter unexpected situations, such as unforeseen obstacles. The Socratic Planner then adjusts plans based on dense visual feedback through a visually-grounded re-planning mechanism. Experiments demonstrate the effectiveness of the Socratic Planner, outperforming current state-of-the-art planning models on the ALFRED benchmark across all metrics, particularly excelling in long-horizon tasks that demand complex inference. We further demonstrate its real-world applicability through deployment on a physical robot for long-horizon tasks.

cs.AI↗

arXiv2025

Non-degenerate noise-resilient superconducting qubit

We propose a superconducting qubit based on engineering the first and second harmonics of the Josephson energy and phase relation $E_{J1}\cos φ$ and $E_{J2}\cos 2φ$. By constructing a circuit such that $E_{J2}$ is negative and $|E_{J1}| \ll |E_{J2}|$, we create a periodic potential with two non-degenerate minima. The qubit, which we dub "harmonium", is formed from the lowest-energy states of each minimum. Bit-flip protection of the qubit arises due to the localization of each qubit state to their respective minima, while phase-flip protection can be understood by considering the circuit within the Born-Oppenheimer approximation. We demonstrate with time-domain simulations that single- and two-qubit gates can be performed in approximately one hundred nanoseconds. Finally, we compute the qubit coherence times using numerical diagonalization of the complete circuit in conjunction with state-of-the-art noise models. We estimate out-of-manifold heating times on the order of milliseconds, which can be treated as erasure errors using conventional dispersive readout. We estimate pure-dephasing times on the order of many tens of milliseconds, and bit-flip times on the order of seconds.

quant-ph↗

arXiv2025

Distinct Optical Excitation Mechanisms of a Coherent Magnon in a van der Waals Antiferromagnet

The control of antiferromagnets with ultrashort optical pulses has emerged as a prominent field of research. Tailored laser excitation can launch coherent spin waves at terahertz frequencies, yet a comprehensive description of their generation mechanisms is still lacking despite extensive efforts. Using terahertz emission spectroscopy, we investigate the generation of a coherent magnon mode in the van der Waals antiferromagnet NiPS$_3$ under a range of photoexcitation conditions. By tuning the pump photon energy from transparency to resonant with a $d$-$d$ transition, we reveal a striking change in the coherent magnon's dependence on the pump polarization, indicating two distinct excitation mechanisms. Our findings provide a strategy for the manipulation of magnetic modes via photoexcitation around sub-gap electronic states.

cond-mat.str-el↗

arXiv2025

Highly efficient nonvolatile magnetization switching and multi-level states by current in single van der Waals topological ferromagnet Fe3GeTe2

Robust multi-level spin memory with the ability to write information electrically is a long-sought capability in spintronics, with great promise for applications. Here we achieve nonvolatile and highly energy-efficient magnetization switching in a single-material device formed of van-der-Waals topological ferromagnet Fe3GeTe2, whose magnetic information can be readily controlled by a tiny current. Furthermore, the switching current density and power dissipation are about 400 and 4000 times smaller than those of the existing spin-orbit-torque magnetic random access memory based on conventional magnet/heavy-metal systems. Most importantly, we also demonstrate multi-level states, switched by electrical current, which can dramatically enhance the information capacity density and reduce computing costs. Thus, our observations combine both high energy efficiency and large information capacity density in one device, showcasing the potential applications of the emerging field of van-der-Waals magnets in the field of spin memory and spintronics.

cond-mat.mtrl-sci↗

arXiv2025

Gigantic current control of coercive field and magnetic memory based on nm-thin ferromagnetic van der Waals Fe3GeTe2

Controlling magnetic states by a small current is essential for the next-generation of energy-efficient spintronic devices. However, it invariably requires considerable energy to change a magnetic ground state of intrinsically quantum nature governed by fundamental Hamiltonian, once stabilized below a phase transition temperature. We report that surprisingly an in-plane current can tune the magnetic state of nm-thin van der Waals ferromagnet Fe3GeTe2 from a hard magnetic state to a soft magnetic state. It is the direct demonstration of the current-induced substantial reduction of the coercive field. This surprising finding is possible because the in-plane current produces a highly unusual type of gigantic spin-orbit torque for Fe3GeTe2. And we further demonstrate a working model of a new nonvolatile magnetic memory based on the principle of our discovery in Fe3GeTe2, controlled by a tiny current. Our findings open up a new window of exciting opportunities for magnetic van der Waals materials with potentially huge impacts on the future development of spintronic and magnetic memory.

cond-mat.mtrl-sci↗

arXiv2024

CLIP-RT: Learning Language-Conditioned Robotic Policies from Natural Language Supervision

Teaching robots desired skills in real-world environments remains challenging, especially for non-experts. A key bottleneck is that collecting robotic data often requires expertise or specialized hardware, limiting accessibility and scalability. We posit that natural language offers an intuitive and accessible interface for robot learning. To this end, we study two aspects: (1) enabling non-experts to collect robotic data through natural language supervision (e.g., "move the arm to the right") and (2) training robot policies directly from this supervision. Specifically, we introduce a data collection framework that collects robot demonstrations based on natural language supervision and further augments these demonstrations. We then present CLIP-RT, a new vision-language-action (VLA) model that learns language-conditioned visuomotor policies from this supervision. CLIP-RT adapts the pretrained CLIP model and learns to predict language-based motion primitives via contrastive imitation learning. We train CLIP-RT on the Open X-Embodiment dataset and finetune it on in-domain data collected by our framework. In real-world evaluations, CLIP-RT demonstrates strong capabilities in learning novel manipulation skills, outperforming OpenVLA (7B parameters) by 24% in average success rates, while using 7x fewer parameters (1B). We further assess CLIP-RT's capabilities in few-shot generalization and collaborative scenarios involving large pretrained models or humans. In simulated environments, CLIP-RT also yields strong performance, achieving a 93.1% average success rate on the LIBERO benchmark with an inference throughput of 163 Hz.

cs.RO↗