He leads the System INtelligence Group (SING).

Root Completion from Intraoral Scans of Tooth Crowns using Diffusion with Patch Perturbation

ContextFace: Generating Facial Expressions from Emotional Contexts

Diagnosis of carpal tunnel syndrome using deep learning with comparative guidance

DuoDent: Tooth Generation using Dual-Stream Diffusion with Normal Consistency

DynASyn: Multi-subject personalization enabling dynamic action synthesis

Syntriever: How to train your retriever with synthetic data from LLMs

Joint Device Selection and Bandwidth Allocation for Layerwise Federated Learning

QPaug: Question and passage augmentation for open-domain question answering of LLMs

Best of Both Modalities: Fusing CBCT and Intraoral Scan Data Into a Single Tooth Image

Hierarchical Position Embedding of Graphs with Landmarks and Clustering for Link Prediction

NeBLa: Neural beer-lambert for 3D reconstruction of oral structures from panoramic radiographs

Practical and efficient coded transmission for full-duplex relay networks without CSI

Automatic assessment of upper extremity function and mobile application for self-administered stroke rehabilitation

Optimal Scheduling for Uncoded and Coded Multicast in Millimeter Wave Networks Leveraging Directionality and Reflections

Distinguishing lumbar radiculopathy using AI-assisted quantitative muscle ultrasound-pilot study

Dual domain diffusion guidance for 3D CBCT metal artifact reduction

Quantitative analysis of temporal parameters correlated with aspiration and lesion location in stroke patients

Application of deep learning technology for temporal analysis of videofluoroscopic swallowing studies

Cemb-sam: Segment anything model with condition embedding for joint learning from heterogeneous datasets

Automatic segmentation of internal tooth structure from CBCT images using hierarchical deep learning

3D teeth reconstruction from panoramic radiographs using neural implicit functions

Contrastive learning for action assessment using graph convolutional networks with augmented virtual joints

Feasibility of artificial intelligence assisted quantitative muscle ultrasound in carpal tunnel syndrome

Federated Semi-Supervised Segmentation with Randomized Weight Perturbation

The reliability and validity of the Korean version of the reaching performance scale for stroke after translation and cross-cultural adaptation

Optimal multicast scheduling for switched beamforming systems leveraging reflections

Comdense: combined dense embedding of relation-aware and common features for knowledge graph completion

Channel-aware scheduling for coded packet broadcasting in full-duplex relay networks

Scale-attentional U-Net for the segmentation of the median nerve in ultrasound images

Keyword Spotting with Synthetic Data using Heterogeneous Knowledge Distillation

Hopfield-type neural ordinary differential equation for robust machine learning

Optimal multicast scheduling for millimeter wave networks leveraging directionality and reflections

Low-complexity coded transmission without CSI for full-duplex relay networks

Real-Time Inter-Vehicle Data Fusion based on A New Metric for Evidence Distance in Autonomous Vehicle Systems

Low-Complexity Coded Transmission Without CSI for Full-Duplex Relay Networks

Blockchain of Finite-Lifetime Blocks with Applications to Edge-based IoT

Cooperative Buffer-Aided Relaying Using Full-Duplex Relays With Flow Control

Statistical Multiplexing and Traffic Shaping Games for Network Slicing

A Robust Proposal Generation Method for Text Lines in Natural Scene Images

Joint Load Balancing and Energy Saving Algorithm for Virtual Network Embedding in Infrastructure Providers

Multicast Scheduling for Relay-based Heterogeneous Networks using Rateless Codes

Delay-Optimal Scheduling for Two-Hop Relay Networks with Randomly Varying Connectivity: Join the Shortest Queue-Longest Connected Queue Policy

Joint routing and scheduling for data collection with compressive sensing to achieve order-optimal latency

Energy-Efficient Collection of Sparse Data in Wireless Sensor Networks Using Sparse Random Matrices

Distributed Power Allocation for Femtocell Networks Subject to Macrocell SINR Balancing

Opportunistic Scheduling of Randomly Coded Multicast Transmissions at Half-Duplex Relay Stations

Application of precise indoor position tracking to immersive virtual reality with translational movement support

Minimum Cost Data Aggregation for Wireless Sensor Networks Computing Functions of Sensed Data

Securing one-way hash chain based incentive mechanism for vehicular ad hoc networks

A Highly Parallelized Decoder for Random Network Coding leveraging GPGPU

Sufficient Conditions on Stable Recovery of Sparse Signals With Partial Support Information

Integrating Random Network Coding with On-Demand Multicast Routing Protocol

Opportunistic Feedback and Scheduling to Reduce Packet Delays in Heterogeneous Wireless Systems

Reducing Feedback Overhead in  Opportunistic Scheduling of Wireless Networks Exploiting Overhearing

Delay-Optimal Opportunistic Scheduling and Approximations: The Log Rule

Retrieval of the Extreme Values under Deadline Constraints in Wireless Sensor Networks

An ultraviolet-curable mold for sub-100-nm lithography

Minimizing energy consumption in large-scale sensor networks through distributed data compression and hierarchical aggregation

Delay-optimal opportunistic scheduling and approximations: The log rule

Unconventional patterning with a modulus-tunable mold: from imprinting to microcontact printing

Enhanced methane hydrate formation with cyclopentane hydrate seeds

Blockchain of finite-lifetime blocks with applications to edge-based IoT

Spatial energy balancing through proactive multipath routing in wireless multihop networks

Statistical multiplexing and traffic shaping games for network slicing

Observation of High‐Aspect‐Ratio Nanostructures Using Capillary Lithography

Effect of hydrophobic silica nanoparticles on the kinetics of methane hydrate formation in water-in-oil emulsions

Hydrophobic particle effects on hydrate crystal growth at the water–oil interface

A linear InGaP/GaAs HBT power amplifier using parallel-combined transistors with IMD3 cancellation

Anti-adhesive behaviors between solid hydrate and liquid aqueous phase induced by hydrophobic silica nanoparticles

Inhibition effects of activated carbon particles on gas hydrate formation at oil–water interfaces

Equilibria of cyclopentane hydrates with varying HLB numbers of sorbitan monoesters in water-in-oil emulsions

Sufficient conditions on stable recovery of sparse signals with partial support information

Effects of salt on the crystal growth and adhesion force of clathrate hydrates

Hydrate seeding effect on the metastability of CH 4 hydrate

Seung Jun Baek

We describe a novel UV-curable mold that is stiff enough for replicating dense sub-100-nm features even with a high aspect ratio. It also allows for flexibility when the mold is prepared on a flexible support such that large area replication can be accomplished. The composite material of the mold is inert to chemicals and solvents. The surface energy is made low with a small amount of releasing agent such that the mold can be removed easily and cleanly after patterning. In addition, the material allows self-replication of the mold. These unique features of the mold material should make the mold quite useful for various patterning purposes.

In this paper, we study how to reduce energy consumption in large-scale sensor networks, which systematically sample a spatio-temporal field. We begin by formulating a distributed compression problem subject to aggregation (energy) costs to a single sink. We show that the optimal solution is greedy and based on ordering sensors according to their aggregation costs-typically related to proximity-and, perhaps surprisingly, it is independent of the distribution of data sources. Next, we consider a simplified hierarchical model for a sensor network including multiple sinks, compressors/aggregation nodes, and sensors. Using a reasonable metric for energy cost, we show that the optimal organization of devices is associated with a Johnson-Mehl tessellation induced by their locations. Drawing on techniques from stochastic geometry, we analyze the energy savings that optimal hierarchies provide relative to previously …

This paper considers the design of multiuser opportunistic packet schedulers for users sharing a time-varying wireless channel from performance and robustness points of view. For a simplified model falling in the classical Markov decision process framework, we numerically compute and characterize mean-delay-optimal scheduling policies. The computed policies exhibit radial sum-rate monotonicity: As users' queues grow linearly, the scheduler allocates service in a manner that deemphasizes the balancing of unequal queues in favor of maximizing current system throughput (being opportunistic). This is in sharp contrast to previously proposed throughput-optimal policies, e.g., Exp rule and MaxWeight (with any positive exponent of queue length). In order to meet performance and robustness objectives, we propose a new class of policies, called the Log rule, that are radial sum-rate monotone (RSM) and provably …

Modulus-tunable ultraviolet (UV) curing molds allow not only physical patterning of imprinting but also chemical patterning of microcontact printing. The mechanical properties of the mold can be tailored by the chain length of an acrylate modulator in the cross-linking reaction. This tunability can be utilized to obtain a proper balance that is needed for a given patterning technique between the rigidity requirement (tensile modulus = 320 MPa) of a mold for patterning a fine structure and the flexibility requirement (tensile modulus = 19.8 MPa) for a conformal contact. Successful low-pressure imprinting of sub-100-nm patterns and microcontact printing of submicrometer patterns verify the suitability of the mold for a wide spectrum of patterning techniques.

This work presents the kinetic promotion effect of sII cyclopentane (CP) hydrate seeds on sI CH 4 hydrate formation in sodium dodecyl sulfate (SDS) solutions. Gas hydrates are one of the promising materials for natural gas storage due to their high volumetric capacity and safety aspect. Generally, a crystal layer of gas hydrate is preferentially created on the surface of an aqueous phase which results in both low hydrate conversion and slow reaction without mechanical agitations. This study introduces CP hydrate seeds at very small quantities to accelerate the nucleation and growth of the CH 4 hydrate with SDS. To choose the appropriate amount of CP in the system for massive sI CH 4 hydrate formation, several hydrate slurries with various amounts of CP were characterized by powder X-ray diffraction and Raman spectroscopy. Additionally, from the visual observations of the hydrate growth behavior, the direction of …

Edge computing is a promising approach for provisioning distributed cloud services to Internet of Things (IoT) systems. Many recent studies propose that edge nodes use blockchain for the decentralized management and access control of IoT data. However, due to the massive volume of data and related transactions, edge servers will eventually run out of space to store the full chain. We introduce scalable and lightweight architecture called LiTiChain, a blockchain of blocks with finite lifetime. In LiTiChain, outdated transactions and blocks, that is, the blocks whose lifetimes are expired, can be safely removed from the chain. Two graphs are merged into the structure of LiTiChain: 1) a tree representing the order of expiry of lifetimes and 2) a linear graph representing the order of block creation. We show that this construction not only ensures the connectivity of the chain after block deletions but also helps to maintain …

In this paper, we investigate the use of proactive multipath routing to achieve energy-efficient operation of ad hoc wireless networks. The focus is on optimizing tradeoffs between the energy cost of spreading traffic and the improved spatial balance of energy burdens. We propose a simple scheme for multipath routing based on spatial relationships among nodes. Then, combining stochastic geometric and queueing models, we develop a continuum model for such networks, permitting an evaluation of different types of scenarios, i.e., with and without energy replenishing and storage capabilities. We propose a parameterized family of energy balancing strategies and study the spatial distributions of energy burdens based on their associated second-order statistics. Our analysis and simulations show the fundamental importance of the tradeoff explored in this paper, and how its optimization depends on the relative …

Next-generation wireless architectures are expected to enable slices of shared wireless infrastructure, which are customized to specific mobile operators/services. Given infrastructure costs and the stochastic nature of mobile services' spatial loads, it is highly desirable to achieve efficient statistical multiplexing among such slices. We study a simple dynamic resource sharing policy, which allocates a “share” of a pool of (distributed) resources to each slice-share constrained proportionally fair (SCPF). We give a characterization of SCPF's performance gains over static slicing and general processor sharing. We show that higher gains are obtained when a slice's spatial load is more “imbalanced” than, and/or “orthogonal” to, the aggregate network load, and that the overall gain across slices is positive. We then address the associated dimensioning problem. Under SCPF, traditional network dimensioning translates to a …

The fabrication of periodic nanostructures is of great interest because of their potential applications in photonic crystals,[1±3] data storage,[4±6] and nanometer-scale biological sensors.[7, 8] To enable such applications, unconventional methods that enable the fabrication of dense, large-area, periodic nanostructures with reasonable control over size and periodicity are required, since conventional photolithography and electron-beam lithography do not appear to be economically viable techniques.[9] Nano-imprint lithography and soft lithography are alternative, cost-effective, high-throughput lithographic techniques that could potentially replace photolithography for nanostructure fabrication in the sub-100 nm size rØgime.[10, 11] In such parallel processes, various patterns are simultaneously formed by the physical contact of a hard or soft mold with a targeted substrate. Recently, by simply combining essential …

This study addresses the effect of hydrophobic silica nanoparticles (SiNPs) on methane hydrate formation in water-in-decane emulsions. Even though the new concept of hydrophobic particle inhibitors has been proposed in the flat interface of oil and bulk water phases, the mechanism of the particle inhibition has not been investigated for a hydrate-forming emulsion system. Numerous experiments with various amounts of SiNPs applied to water-in-decane emulsions and different temperature and pressure profiles have verified how the nanoparticles interact at the interface for hydrate inhibition or promotion. At lower concentrations of SiNPs, they act as a hydrate inhibitor. However, as the amount of SiNPs increases beyond a certain concentration (0.5 wt %), the formation rate anomalously increases because the oil film of the emulsions containing methane becomes thicker with the increasing amount of SiNPs, and …

This study introduced hydrophobic silica nanoparticles (SiNPs) into an interface of aqueous and hydrate‐forming oil phases and analyzed the inhibition of hydrate crystal growth after seeding the hydrate slurry. The hydrate inhibition performance was quantitatively identified by micro‐differential scanning calorimetry (micro‐DSC) experiments. Through the addition of 1.0 wt % of SiNPs into the water–oil interface, the hydrate crystal growth only occurred around the seeding position of cyclopentane (CP) hydrate slurry, and the growth of hydrate crystals was retarded. Upon a further increase in the SiNP concentration up to 2.0 wt %, the SiNP‐laden interface completely prevented hydrate growth. We observed a hollow conical shape of hydrate crystals with 0.0 and 1.0 wt % of SiNPs, respectively, but the size and shape of the conical crystals was shrunken at 1.0 wt % of silica nanoparticles. However, the conical …

In this letter, we present a linear InGaP/GaAs HBT power amplifier (PA) with parallel-combined transistors for small-cell applications. Ballast resistors with bypass resistors and capacitors are employed in parallel-combined transistors. When the resistors and capacitors are set to appropriate values, IMD3 components of the parallel-combined transistors are found to be out of phase with each other by 180° and are canceled out at the output. The experimental results show that the proposed HBT PA with parallel-combined transistors produces a saturated output power of 33.5 dBm at 0.88 GHz, with a power-added efficiency (PAE) of 46.1% at a 5-V supply voltage. To validate the effectiveness of the proposed HBT PA for linearity improvement, the implemented PA is also tested with a long-term evolution (LTE) signal (8.1-dB PAPR with 10-MHz bandwidth). The proposed PA achieves an adjacent channel leakage ratio …

This study introduces an “anti-adhesive force” at the interface of solid hydrate and liquid solution phases. The force was induced by the presence of hydrophobic silica nanoparticles or one of the common anti-agglomerants (AAs), sorbitan monolaurate (Span 20), at the interface. The anti-adhesive force, which is defined as the maximum pushing force that does not induce the formation of a capillary bridge between the cyclopentane (CP) hydrate particle and the aqueous solution, was measured using a microbalance. Both hydrophobic silica nanoparticles and Span 20 can inhibit adhesion between the CP hydrate probe and the aqueous phase because silica nanoparticles have an aggregative property at the interface, and Span 20 enables the hydrate surface to be wetted with oil. Adding water-soluble sodium dodecyl sulfate (SDS) to the nanoparticle system cannot affect the aggregative property or the distribution …

This study presents the inhibition effects of activated carbon particles on the formation of cyclopentane and propane mixed hydrates in the presence of non-hydrate formers and a surfactant. Activated carbon particles gather around the interface between water and oil phases due to their hydrophobicity and this particle layer prevents water molecules from contacting any hydrate guests. The coexistence of cyclopentane and propane mixed hydrates was clearly observed through Raman spectroscopy. The Raman spectra showed that the propane and cyclopentane were enclathrated in the large cages of structure II hydrate. The inhibition efficiency of the activated carbon particles was quantified using high pressure micro differential scanning calorimetry (HP Micro-DSC). The dissociation enthalpies of the mixed hydrates were significantly reduced at 0.5 wt% of activated carbon particles based on the weight of the oil …

This study investigated the effect of hydrophilic-lipophilic balance (HLB) numbers of sorbitan monoesters on equilibrium conditions of cyclopentane hydrates in water-in-oil emulsions with or without NaCl. HLB numbers of the surfactants (Span 20, 40, 60, and 80), which have the same hydrophilic group, are determined by the length and shape of alkyl chain and the HLB number refers to hydrophobicity of surfactant. The equilibrium state of the hydrate phase can be affected by the HLB of the surfactants whose number is the lowest in Span 80 (4.3) and the highest in Span 20 (8.6). The emulsion size is proportional to the increasing HLB number of sorbitan monoesters. Equilibrium dissociation temperatures of cyclopentane hydrate in the water-in-oil emulsion also rise as the HLB number increases, which was determined in micro differential scanning calorimetry (micro-DSC). In the presence of 3.5 wt % NaCl in the …

In this study, we propose an application for immersive virtual reality experiences, which integrates three-dimensional (3D) head-mounted displays (HMDs) with a precise indoor position tracking algorithm based on ultrasound. Our method provides a natural virtual reality experience with interaction by precisely matching the physical movements in the real world with those in the virtual environment, unlike other methods that require external input devices to move around in the virtual environment. Users can move within the assigned indoor space while carrying a wireless client device with the HMD, without the risk of colliding with obstacles or structures. The system is designed to provide the accurate 3D X, Y, and Z coordinate values of translational movements in real-time as well as the pitch, roll, and yaw values of rotational movements supported by the HMD, resulting in the six degrees of freedom required by …

In this letter, we study signal reconstruction from compressed sensing measurements. We propose new sufficient conditions for stable recovery when partial support information is available. Weighted -minimization is adopted to recover the original signal under three noise models. The proposed approach is to use Ozeki's inequality and shifting inequality in order to bound the errors in the associated weighted -minimization. Our result offers generalized performance bounds on recovery capturing known support information. Improved sufficient conditions for recovery are derived based on our results, even for the cases where the accuracy of prior support information is arbitrarily low.

In the offshore oil and gas developments, the problems in relation to hydrate flow assurance are always influenced by the existence of salt in the pipeline systems. Even though the salt effects are essential factors associated with the adhesion force between clathrate hydrate and aqueous solutions, the roles of salt in hydrate adhesion behaviors have rarely been considered. This study aims at understanding the effects of salt on not only the adhesion force variations but also the transformation of hydrate growth morphology. The presence of salt did not profoundly influence the change in the adhesion force of the hydrate probe with surfactant-free solutions, whereas it drastically reduced the adhesion force of the hydrate system with surfactant-dissolving aqueous solutions. The experimental results also revealed the inhibition effect of salt on the hydrate crystal growth through the observation of microscopic images …

Temporal parameters during swallowing are analyzed for objective and quantitative evaluation of videofluoroscopic swallowing studies (VFSS). Manual analysis by clinicians is time-consuming, complicated and prone to human error during interpretation; therefore, automated analysis using deep learning has been attempted. We aimed to develop a model for the automatic measurement of various temporal parameters of swallowing using deep learning. Overall, 547 VFSS video clips were included. Seven temporal parameters were manually measured by two physiatrists as ground-truth data: oral phase duration, pharyngeal delay time, pharyngeal response time, pharyngeal transit time, laryngeal vestibule closure reaction time, laryngeal vestibule closure duration, and upper esophageal sphincter opening duration. ResNet3D was selected as the base model for the deep learning of temporal parameters. The …

Cyclopentane (CP) hydrate seeds can lead to nucleation of CH 4 hydrate with a lower supersaturation; the concept of nucleation potential was applied to estimate the metastable zone width (MSZW) of CH 4 hydrate. To verify the crystal structure of CH 4 hydrate formed from the CP hydrate seeds, the hydrate samples were analyzed by high resolution powder diffraction (HRPD). 1 wt% of CP hydrates in the system reduced the MSZW of CH 4 hydrate from 3.39 K to 1.32 K, and showed synergetic performance with sodium dodecyl sulfate (SDS). From the hydrate nucleation theory, SDS is able to decrease the effective surface energy for heterogeneous nucleation on the stainless steel wall, but the CP hydrate seeds provide new nucleation sites with even lower surface energy than that of the stainless steel wall. Hence, the nucleation rate depends on the amount of CP hydrate seeds, and the kinetic parameter can be …

System INtelligence Group (지능시스템 연구실)

교수님 소개

백승준

연구실 소개

연구실 키워드

알럼나이가 간 회사

연구실 지원 방법

대우 조건

연도별 논문 인용수

저널 & 학회 논문 출판 수

연구실 논문 목록

Root Completion from Intraoral Scans of Tooth Crowns using Diffusion with Patch Perturbation

ContextFace: Generating Facial Expressions from Emotional Contexts

DuoDent: Tooth Generation using Dual-Stream Diffusion with Normal Consistency

연구실 보유 특허

대학별 입시끈

우리 연구실 모집

교수님 소개

백승준

연구실 키워드

알럼나이가 간 회사

연구실 지원 방법

대우 조건

연도별 논문 인용수

저널 & 학회 논문 출판 수

연구실 논문 목록

Root Completion from Intraoral Scans of Tooth Crowns using Diffusion with Patch Perturbation

ContextFace: Generating Facial Expressions from Emotional Contexts

DuoDent: Tooth Generation using Dual-Stream Diffusion with Normal Consistency

연구실 보유 특허

최근 3년간 연구실적 높은 연구실

대학별 입시끈

우리 연구실 모집