arXiv subjects

Sangmin Lee

Publications and source records attributed to Sangmin Lee.

At least 91 records · Page 5Linked to original sources

arXiv2021

Observation of metallic electronic structure in a single-atomic-layer oxide

Correlated electrons in transition metal oxides (TMOs) exhibit a variety of emergent phases. When TMOs are confined to a single-atomic-layer thickness, experiments so far have shown that they usually lose diverse properties and become insulators. In an attempt to extend the range of electronic phases of the single-atomic-layer oxide, we search for a metallic phase in a monolayer-thick epitaxial SrRuO$_3$ film. Combining atomic-scale epitaxy and angle-resolved photoemission measurements, we show that the monolayer SrRuO$_3$ is a strongly correlated metal. Systematic investigation reveals that the interplay between dimensionality and electronic correlation makes the monolayer SrRuO$_3$ an incoherent metal with orbital-selective correlation. Furthermore, the unique electronic phase of the monolayer SrRuO$_3$ is found to be highly tunable, as charge modulation demonstrates an incoherent-to-coherent crossover of the two-dimensional metal. Our work emphasizes the potentially rich phases of single-atomic-layer oxides and provides a guide to the manipulation of their two-dimensional correlated electron systems.

cond-mat.str-el↗

arXiv2021

The Relativistic Spherical Top as a Massive Twistor

We prove the equivalence between two traditional approaches to the classical mechanics of a massive spinning particle in special relativity. One is the spherical top model of Hanson and Regge, recast in a Hamiltonian formulation with improved treatment of covariant spin constraints. The other is the massive twistor model, slightly generalized to incorporate the Regge trajectory relating the mass to the total spin angular momentum. We establish the equivalence by computing the Dirac brackets of the physical phase space carrying three translation and three rotation degrees of freedom. Lorentz covariance and little group covariance uniquely determine the structure of the physical phase space. The Regge trajectory does not affect the phase space but enters the equations of motion. Upon quantization, the twistor model produces a spectrum that agrees perfectly with the massive spinor-helicity description proposed by Arkani-Hamed, Huang and Huang for scattering amplitudes for all masses and spins.

hep-th↗

arXiv2021

Two-stage Textual Knowledge Distillation for End-to-End Spoken Language Understanding

End-to-end approaches open a new way for more accurate and efficient spoken language understanding (SLU) systems by alleviating the drawbacks of traditional pipeline systems. Previous works exploit textual information for an SLU model via pre-training with automatic speech recognition or fine-tuning with knowledge distillation. To utilize textual information more effectively, this work proposes a two-stage textual knowledge distillation method that matches utterance-level representations and predicted logits of two modalities during pre-training and fine-tuning, sequentially. We use vq-wav2vec BERT as a speech encoder because it captures general and rich features. Furthermore, we improve the performance, especially in a low-resource scenario, with data augmentation methods by randomly masking spans of discrete audio tokens and contextualized hidden representations. Consequently, we push the state-of-the-art on the Fluent Speech Commands, achieving 99.7% test accuracy in the full dataset setting and 99.5% in the 10% subset setting. Throughout the ablation studies, we empirically verify that all used methods are crucial to the final performance, providing the best practice for spoken language understanding. Code is available at https://github.com/clovaai/textual-kd-slu.

cs.CL↗

arXiv2021

Towards a Better Understanding of VR Sickness: Physical Symptom Prediction for VR Contents

We address the black-box issue of VR sickness assessment (VRSA) by evaluating the level of physical symptoms of VR sickness. For the VR contents inducing the similar VR sickness level, the physical symptoms can vary depending on the characteristics of the contents. Most of existing VRSA methods focused on assessing the overall VR sickness score. To make better understanding of VR sickness, it is required to predict and provide the level of major symptoms of VR sickness rather than overall degree of VR sickness. In this paper, we predict the degrees of main physical symptoms affecting the overall degree of VR sickness, which are disorientation, nausea, and oculomotor. In addition, we introduce a new large-scale dataset for VRSA including 360 videos with various frame rates, physiological signals, and subjective scores. On VRSA benchmark and our newly collected dataset, our approach shows a potential to not only achieve the highest correlation with subjective scores, but also to better understand which symptoms are the main causes of VR sickness.

cs.CV↗

arXiv2021

Visual Comfort Aware-Reinforcement Learning for Depth Adjustment of Stereoscopic 3D Images

Depth adjustment aims to enhance the visual experience of stereoscopic 3D (S3D) images, which accompanied with improving visual comfort and depth perception. For a human expert, the depth adjustment procedure is a sequence of iterative decision making. The human expert iteratively adjusts the depth until he is satisfied with the both levels of visual comfort and the perceived depth. In this work, we present a novel deep reinforcement learning (DRL)-based approach for depth adjustment named VCA-RL (Visual Comfort Aware Reinforcement Learning) to explicitly model human sequential decision making in depth editing operations. We formulate the depth adjustment process as a Markov decision process where actions are defined as camera movement operations to control the distance between the left and right cameras. Our agent is trained based on the guidance of an objective visual comfort assessment metric to learn the optimal sequence of camera movement actions in terms of perceptual aspects in stereoscopic viewing. With extensive experiments and user studies, we show the effectiveness of our VCA-RL model on three different S3D databases.

cs.CV↗

arXiv2021

Newtonian Event-Chain Monte Carlo and Collision Prediction with Polyhedral Particles

Polyhedral nanocrystals are building blocks for nanostructured materials that find applications in catalysis and plasmonics. Synthesis efforts and self-assembly experiments have been assisted by computer simulations that predict phase equilibra. Most current simulations employ Monte Carlo methods, which generate stochastic dynamics. Collective and correlated configuration updates are alternatives that promise higher computational efficiency and generate trajectories with realistic dynamics. One such alternative involves event-chain updates and has recently been proposed for spherical particles. In this contribution, we develop and apply event-chain Monte Carlo for hard convex polyhedra. Our simulation makes use of an improved computational geometry algorithm XenoSweep, which predicts sweep collision in a particularly simple way. We implement Newtonian event chains in the open source general-purpose particle simulation toolkit HOOMD-blue for serial and parallel simulation. The speed-up over state-of-the-art Monte Carlo is between a factor of 10 for nearly spherical polyhedra and a factor of 2 for highly aspherical polyhedra. Finally, we validate the Newtonian event-chain algorithm by applying it to a current research problem, the multi-step nucleation of two classes of hard polyhedra.

cond-mat.stat-mech↗

arXiv2021

Video Prediction Recalling Long-term Motion Context via Memory Alignment Learning

Our work addresses long-term motion context issues for predicting future frames. To predict the future precisely, it is required to capture which long-term motion context (e.g., walking or running) the input motion (e.g., leg movement) belongs to. The bottlenecks arising when dealing with the long-term motion context are: (i) how to predict the long-term motion context naturally matching input sequences with limited dynamics, (ii) how to predict the long-term motion context with high-dimensionality (e.g., complex motion). To address the issues, we propose novel motion context-aware video prediction. To solve the bottleneck (i), we introduce a long-term motion context memory (LMC-Memory) with memory alignment learning. The proposed memory alignment learning enables to store long-term motion contexts into the memory and to match them with sequences including limited dynamics. As a result, the long-term context can be recalled from the limited input sequence. In addition, to resolve the bottleneck (ii), we propose memory query decomposition to store local motion context (i.e., low-dimensional dynamics) and recall the suitable local context for each local part of the input individually. It enables to boost the alignment effects of the memory. Experimental results show that the proposed method outperforms other sophisticated RNN-based methods, especially in long-term condition. Further, we validate the effectiveness of the proposed network designs by conducting ablation studies and memory feature analysis. The source code of this work is available.

cs.CV↗

arXiv2021

Importance of anisotropic Coulomb interactions in the electronic and magnetic properties of Mn$_3$O$_4$

We report the importance of anisotropic Coulomb interactions in DFT+U calculations of the electronic and magnetic properties of Mn$_3$O$_4$. The effects of anisotropic interactions in Mn$^{2+}$ and Mn$^{3+}$ are separately examined by defining two different sets of Hubbard parameters: $U^{2+}$ and $J^{2+}$ for Mn$^{2+}$ and $U^{3+}$ and $J^{3+}$ for Mn$^{3+}$. The anisotropic interactions in Mn$^{3+}$ have a significant impact on the physical properties of Mn$_3$O$_4$ including local magnetic moments, canted angle, spontaneous magnetic moment, and superexchange coupling, but those in Mn$^{2+}$ do not make any noticeable difference. Weak ferromagnetic interchain superexchange, observed in experiments, is predicted only if a sizable anisotropic interaction is considered in Mn$^{3+}$. By analyzing the eigenoccupations of the on-site Mn density matrix, we found that the spin channel involving Mn$^{3+}$ $d_{x^2-y^2}$ orbitals, which governs the 90$^\circ$ correlation superexchange, is directly controlled by the anisotropic interactions. These findings demostrate that the exchange correction $J$ for the intraorbital Coulomb potential is of critical importance for first-principles description of reduced Mn oxides containing Mn$^{3+}$ or Mn$^{4+}$.

cond-mat.mtrl-sci↗

arXiv2020

Explaining Convolutional Neural Networks through Attribution-Based Input Sampling and Block-Wise Feature Aggregation

As an emerging field in Machine Learning, Explainable AI (XAI) has been offering remarkable performance in interpreting the decisions made by Convolutional Neural Networks (CNNs). To achieve visual explanations for CNNs, methods based on class activation mapping and randomized input sampling have gained great popularity. However, the attribution methods based on these techniques provide lower resolution and blurry explanation maps that limit their explanation power. To circumvent this issue, visualization based on various layers is sought. In this work, we collect visualization maps from multiple layers of the model based on an attribution-based input sampling technique and aggregate them to reach a fine-grained and complete explanation. We also propose a layer selection strategy that applies to the whole family of CNN-based models, based on which our extraction framework is applied to visualize the last layers of each convolutional block of the model. Moreover, we perform an empirical analysis of the efficacy of derived lower-level information to enhance the represented attributions. Comprehensive experiments conducted on shallow and deep models trained on natural and industrial datasets, using both ground-truth and model-truth based evaluation metrics validate our proposed algorithm by meeting or outperforming the state-of-the-art methods in terms of explanation ability and visual quality, demonstrating that our method shows stability regardless of the size of objects or instances to be explained.

cs.CV↗

arXiv2020

Complete Hamiltonian for spinning binary systems at first post-Minkowskian order

Building upon recent progress in applying on-shell amplitude techniques to classical observables in general relativity, we propose a closed-form formula for the conservative Hamiltonian of a spinning binary system at the 1st post-Minkowskian (1PM) order. It is applicable for general spinning bodies with arbitrary spin multipole moments. The formula is linear in gravitational constant by definition, but exact to all orders in momentum and spin expansions. At each spin order, our formula implies that the spin-dependence and momentum dependence factorize almost completely. We expand our formula in momentum and compare the terms with 1PM parts of the post-Newtonian computations in the literature. Up to canonical transformations, our results agree perfectly with all previous ones. We also compare our formula for black hole to that derived from a spinning test-body near a Kerr black hole via the effective one-body mapping, and find perfect agreement.

hep-th↗

arXiv2020

Effects of paramagnetic fluctuations on the thermochemistry of MnO (100) surfaces in the oxygen evolution reaction

We investigated the effects of paramagnetic (PM) fluctuations on the thermochemistry of the MnO(100) surface in the oxygen evolution reaction (OER) using the "noncollinear magnetic sampling method \textit{plus} $U$" (NCMSM$+U$). Various physical properties, such as the electronic structure, free energy, and charge occupation, of the MnO (100) surface in the PM state with several OER intermediates, were reckoned and compared to those in the antiferromagnetic (AFM) state. We found that PM fluctuation enhances charge transfer from a surface Mn ion to each of the intermediates and strengthens the chemical bond between them, while not altering the overall features, such as the rate determining step and resting state, in reaction pathways. The enhanced charge transfer can be attributed to the delocalized nature of valence bands observed in the PM surface. In addition, it was observed that chemical-bond enhancement depends on the intermediates, resulting in significant deviations in reaction energy barriers. Our study suggests that PM fluctuations play a significant role in the thermochemistry of chemical reactions occurring on correlated oxide surfaces.

cond-mat.mtrl-sci↗

arXiv2020

Comprehensive Facial Expression Synthesis using Human-Interpretable Language

Recent advances in facial expression synthesis have shown promising results using diverse expression representations including facial action units. Facial action units for an elaborate facial expression synthesis need to be intuitively represented for human comprehension, not a numeric categorization of facial action units. To address this issue, we utilize human-friendly approach: use of natural language where language helps human grasp conceptual contexts. In this paper, therefore, we propose a new facial expression synthesis model from language-based facial expression description. Our method can synthesize the facial image with detailed expressions. In addition, effectively embedding language features on facial features, our method can control individual word to handle each part of facial movement. Extensive qualitative and quantitative evaluations were conducted to verify the effectiveness of the natural language.

cs.CV↗

arXiv2019

"Lagrangian Disks" in M-theory

While the study of bordered (pseudo-)holomorphic curves with boundary on Lagrangian submanifolds has a long history, a similar problem that involves (special) Lagrangian submanifolds with boundary on complex surfaces appears to be largely overlooked in both physics and math literature. We relate this problem to geometry of coassociative submanifolds in $G_2$ holonomy spaces and to $Spin(7)$ metrics on 8-manifolds with $T^2$ fibrations. As an application to physics, we propose a large class of brane models in type IIA string theory that generalize brane brick models on the one hand and 2d theories $T[M_4]$ on the other.

hep-th↗

arXiv2019

The simplest massive S-matrix: from minimal coupling to Black Holes

In this paper, we explore the physics of electromagnetically and gravitationally coupled massive higher spin states from the on-shell point of view. Starting with the three-point amplitude, we focus on the simplest amplitude which is characterized by matching to minimal coupling in the UV. In the IR such amplitude leads to g = 2 for arbitrary charged spin states, and the best high energy behavior for a given spin. We proceed to construct the (gravitational) Compton amplitude for generic spins. We find that the leading deformation away from minimal coupling, in the gravitation sector, will lead to inconsistent factorizations and are thus forbidden. As the corresponding deformation in the gauge sector encodes the anomalous magnetic dipole moment, this leads to the prediction that for systems with gauge2 =gravity relations, such as perturbative string theory, all charged states must have g = 2. It is then natural to ask for generic spin, what is the theory that yields such minimal coupling. By matching to the one body effective action, remarkably we verify that for large spins, the answer is Kerr black holes. This identification is then an on-shell avatar of the no hair theorem. Finally using this identification as well as the newly constructed Compton amplitudes, we proceed to compute the spin dependent pieces for the classical potential at 2PM order up to degree four in spin operator of either black holes.

hep-th↗

arXiv2019

Noncollinear magnetic sampling method for paramagnetic Mott insulator MnO

We present a new approach based on the static density functional theory (DFT) to describe paramagentic MnO, which is a representative paramagnetic Mott insulator. We appended the spin noncollinearity and the canonical ensemble to the magnetic sampling method (MSM), which is one of the supercell approaches based on disordered local moment model. The combination of the noncollinear MSM (NCMSM) with DFT$+U$ represents a highly favorable computational method called NCMSM$+U$ to accurately determine the paramagnetic properties of MnO with moderate numerical cost. The effects of electron correlations and spin noncollinearity on the properties of MnO were also investigated. We revealed that the spin noncollinearity plays an important role in determining the detailed electronic profile and precise energetics of paramagnetic MnO. Our results illustrate that the NCMSM$+U$ approach may be used as an alternative to the $\textit{ab initio}$ framework of dynamic mean field theory based on DFT in the simulation of the high-temperature properties of Mott insulators.

cond-mat.str-el↗

arXiv2018

STAN: Spatio-Temporal Adversarial Networks for Abnormal Event Detection

In this paper, we propose a novel abnormal event detection method with spatio-temporal adversarial networks (STAN). We devise a spatio-temporal generator which synthesizes an inter-frame by considering spatio-temporal characteristics with bidirectional ConvLSTM. A proposed spatio-temporal discriminator determines whether an input sequence is real-normal or not with 3D convolutional layers. These two networks are trained in an adversarial way to effectively encode spatio-temporal features of normal patterns. After the learning, the generator and the discriminator can be independently used as detectors, and deviations from the learned normal patterns are detected as abnormalities. Experimental results show that the proposed method achieved competitive performance compared to the state-of-the-art methods. Further, for the interpretation, we visualize the location of abnormal events detected by the proposed networks using a generator loss and discriminator gradients.

cs.CV↗

arXiv2017

Dynamics Transfer GAN: Generating Video by Transferring Arbitrary Temporal Dynamics from a Source Video to a Single Target Image

In this paper, we propose Dynamics Transfer GAN; a new method for generating video sequences based on generative adversarial learning. The spatial constructs of a generated video sequence are acquired from the target image. The dynamics of the generated video sequence are imported from a source video sequence, with arbitrary motion, and imposed onto the target image. To preserve the spatial construct of the target image, the appearance of the source video sequence is suppressed and only the dynamics are obtained before being imposed onto the target image. That is achieved using the proposed appearance suppressed dynamics feature. Moreover, the spatial and temporal consistencies of the generated video sequence are verified via two discriminator networks. One discriminator validates the fidelity of the generated frames appearance, while the other validates the dynamic consistency of the generated video sequence. Experiments have been conducted to verify the quality of the video sequences generated by the proposed method. The results verified that Dynamics Transfer GAN successfully transferred arbitrary dynamics of the source video sequence onto a target image when generating the output video sequence. The experimental results also showed that Dynamics Transfer GAN maintained the spatial constructs (appearance) of the target image while generating spatially and temporally consistent video sequences.

cs.CV↗

arXiv2017

Elliptic Genera of 2d (0,2) Gauge Theories from Brane Brick Models

We compute the elliptic genus of abelian 2d (0,2) gauge theories corresponding to brane brick models. These theories are worldvolume theories on a single D1-brane probing a toric Calabi-Yau 4-fold singularity. We identify a match with the elliptic genus of the non-linear sigma model on the same Calabi-Yau background, which is computed using a new localization formula. The matching implies that the quantum effects do not drastically alter the correspondence between the geometry and the 2d (0,2) gauge theory. In theories whose matter sector suffers from abelian gauge anomaly, we propose an ansatz for an anomaly cancelling term in the integral formula for the elliptic genus. We provide an example in which two brane brick models related to each other by Gadde-Gukov-Putrov triality give the same elliptic genus.

hep-th↗