Daegu Gyeongbuk Institute of Science and Technology
DGIST Library Institutional RepositoryNot a member yet
12664 research outputs found
Sort by
RainSD: Rain style diversification module for image synthesis enhancement using feature-level style distribution
No full textAutonomous driving technology nowadays targets to level 4 or beyond, but the researchers are faced with some limitations for developing reliable driving algorithms in diverse challenges. To promote the spread of autonomous vehicles widely, it is important to address safety issues in this technology. Among various safety concerns, the sensor blockage problem by severe weather conditions can be one of the most frequent threats for multi-task learning-based perception algorithms during autonomous driving. To handle this problem, the importance of generating proper datasets is becoming more significant. In this paper, a synthetic road dataset with sensor blockage generated from real road dataset BDD100K is suggested in the format of BDD100K annotation. Rain streaks for each frame were made using an experimentally established equation and translated utilizing the image-to-image translation network based on style transfer. Using this dataset, the degradation of the diverse multitask networks for autonomous driving, such as lane detection, driving area segmentation, and traffic object detection, has been thoroughly evaluated and analyzed. The tendency of performance degradation of deep neural network-based perception systems for autonomous vehicles has been analyzed in depth. Finally, we discuss the limitation and future directions of deep neural network-based perception algorithms and autonomous driving dataset generation based on image-to-image translation. © 2025FALSEsciescopu
딥 러닝을 위한 에너지 확장형 저전력 추론 하드웨어
No full textConvolutional Neural Network, Processing-In-Memory, Energy-Scalability, SW-HW Co-OptimizationConvolutional Neural Networks (CNNs) have demonstrated an outstanding performance in image analysis by learning from image features. However, as image analysis tasks become more complex and diverse, CNN models continue to grow in both structural complexity and model size. In this thesis, we present an SW- HW co-optimization methodology to enable efficient inference of CNN models, especially on embedded platforms with limited hardware resources. To maximize energy efficiency while maintaining the original model accuracy, we propose a specially designed CNN model that consists of two different bit-precision paths. By adjusting the ratio of low-precision and high-precision paths, the proposed network enables to achieve energy scalability, supporting models from low-power models to high-accuracy but large models. Additionally, we introduce an automatic search algorithm based on an evolutionary algorithm that layer-wisely explores the optimal ratio of the two precision paths to satisfy the specified constraints. For the SW-HW co-optimization, we also propose a novel hardware accelerator architecture that can energy-efficiently perform the inference of the dual-precision network. This employs a heterogeneous Processing-in-Memory (PIM) structure designed with two different types of memory. To further enhance energy efficiency, we developed a simulator that can find out the optimal data mapping when the dual-precision networks are deployed onto the heterogeneous PIM architecture. This simulator generates an optimal data mapping strategy based on various performance metrics (latency, memory footprint, and energy consumption). By integrating these proposed techniques, the experimental results show significant improvements in overall performance metrics, demonstrating the effectiveness of the co-design methodology Additionally, a novel heterogeneous PIM architecture that combines digital-based PIM and analog-based PIM is proposed to overcome the challenges and limitations of analog-based PIM. This approach enables the development of inference accelerators that achieve both high energy efficiency and model accuracy. Through this thesis, we demonstrate the potential and scalability of a heterogeneous PIM architecture capable of effectively executing the inference on various deep learning models.
Keywords: Convolutional Neural Network, Processing-In-Memory, Energy-Scalability, SW-HW Co- Optimization
|본 논문은 이미지들의 인식, 분류, 분석 등에서 뛰어난 성능을 보여주고 있는 딥러닝 기반 컨볼루션 신경망(CNN)의 연산 효율성을 높일 수 있는 새로운 알고리즘 기법을 다룬다. 최근 이미지 분석 작업이 점점 더 복잡하고 다양해짐에 따라 CNN 모델은 구조적 복잡성과 모델 크기 모두에서 계속 성장하고 있다. 이 논문에서는 하드웨어 리소스가 제한된 임베디드 플랫폼을 기반으로, CNN 모델을 효율적으로 추론할 수 있도록 SW-HW 공동 최적화 방법론을 제시힌다. 제안한 CNN 모델은 기존 모델 정확도를 유지하면서 에너지 효율을 극대화하기 위해 서로 다른 두 가지 비트 정밀도 (저정밀 및 고정밀)로 양자화된 경로를 결합한 구조를 가진다. 두 경로의 비율을 조정함으로써, 제안한 CNN 모델은 최소한의 에너지를 가진 소형 모델부터 더 높은 정확도를 가진 에너지 집약적 모델에 이르기까지 다양한 모델을 생성할 수 있는 에너지 확장성을 달성할 수 있다. 또한 유저가 제공하는 다양한 제약 조건들 (에너지 효율성 및 메모리 크기)을 충족하기 위해 두 정밀 경로의 최적 비율을 계층별로 탐색하는 진화 알고리즘 기반 자동 검색 알고리즘을 제안한다. SW-HW 공동 최적화를 위해 이중 정밀 네트워크의 추론을 에너지 효율적으로 수행할 수 있는 새로운 하드웨어 가속기 아키텍처도 제안된다. 이 가속기는 두 가지 유형의 메모리로 설계된 이기종 PIM(Processing-in-Memory) 구조를 사용하게 된다. 에너지 효율성을 더욱 향상시키기 위해 이중 정밀 네트워크가 이기종 PIM 아키텍처에서 연산될 때 최적의 데이터 매핑을 찾을 수 있는 시뮬레이터 또한 설계가 되었다. 이 시뮬레이터는 다양한 성능 지표들 (모델 추론시 필요한 지연 시간, 메모리 크기 및 에너지 소비)을 기반으로 최적의 데이터 매핑 전략을 생성한다. 앞서 제안된 기술들을 통합함으로써 전체 성능 지표에서 상당한 개선을 보여주며, CNN의 실제 배포를 발전시키는 데 있어 이 공동 설계 방법론의 효과를 입증한다. 추가적으로, 아날로그 기반 단일 PIM의 문제점과 한계성을 극복하기 위한 디지털 기반 PIM과 아날로그 기반 PIM을 결합한 새로운 이기종 PIM 구조를 제시함으로써, 높은 에너지 효율성과 모델 정확성을 확보할 수 있는 추론 하드웨어 가속기도 제안된다. 본 논문을 통해, 다양한 딥러닝 모델을 효과적으로 추론할 수 있는 이기종 PIM 구조에 대한 새로운 가능성 및 확장성을 입증하고 이를 제안한다.
핵심어: 컨벌루션 신경망, 메모리 내 프로세싱, 에너지 확장성, SW-HW 공동 최적화List of Contents
Abstract i
List of Contents ii
List of Tables v
List of Figures vii
I. Introduction 1
II. Background on Efficient Deep Learning Inference 6
2.1 Deep Compression methods 7
2.1.1 Pruning 7
2.1.2 Quantization 10
2.1.3 Low-Rank Decomposition 16
2.1.4 Network Architecture Search 17
2.2 Processing-In-Memory 19
III. Energy-Scalable Deep Learning Accelerator based on Mixed-Precision
Quantization 22
3.1 Motivation 22
3.2 Mixed-precision CNN Model 24
3.2.1 Proposed Model Architecture 24
3.2.2 Experimental Results 25
3.3 Hardware Architecture 29
3.3.1 Proposed Reconfigurable Multiplier 29
3.3.2 Hardware Evaluation 32
3.4 Noise Resilience 35
IV. Dual-Precision and Low-Power CNN Inference Engine Using SRAM- and eDRAM-
based Processing-in-Memory Arrays 37
4.1 PIM-aware CNN Model 37
4.1.1 Proposed Model Architecture 38
4.1.2 Experimental Results 40
4.2 Heterogeneous PIM Architecture 40
4.2.1 Motivation 40
4.2.2 Proposed PIM Architecture 42
4.2.3 Operations of PIM Arrays 45
4.2.4 Hardware Evaluation 52
4.3 Discussion 55
V. Software-to-Hardware Co-Optimization Methodology for Highly-Efficient Deep
Learning Inference 56
5.1 SW-HW Co-Optimization Methodology 56
5.2 Training Methods for Energy-Scalable CNN Models 57
5.3 Automated Search Algorithm 65
5.4 Experimental Results 70
5.5 Data Mapping Algorithm 74
5.6 Hardware Evaluation 77
5.7 Discussions 80
VI. CAM-CIM: A Hybrid Compute-in-Memory Using Content-Addressable Memory
with Subword Split Mapping for Reduced ADC Resolution 83
6.1 Proposed Hybrid CAM-CIM Architecture 84
6.2 Details of CAM Architecture 85
6.3 Details of CIM Architecture 90
6.4 Experimental Results 93
6.5 Hardware Evaluation 95
VII. Conclusion 97
References 99DoctordCollectio
Large-Scale Molybdenum Disulfide for Hybrid Optoelectronic Devices
No full textTwo-dimensional Material, Quantum Dot, Nanostructure, Transition Metal Dichalcogenide, Hybrid PhotodetectorABSTRACT i
List of Contents ii
List of Tables and Figures v
1. INTRODUCTION 1
1.1 Basic Theory of Low Dimensional Materials 1
1.1.1 Two-Dimensional(2D) Materials 3
1.1.2 Zero-Dimensional(0D) Materials 5
1.2 Low-Dimensional Transistor and Photodetector 8
1.2.1 Low-Dimensional Transistor 8
1.2.2 Low-Dimensional Photodetector 9
1.3 Hybrid Structure of Zero-Dimensional and Two-Dimensional with Application 12
1.3.1 0D and 2D hybrid Structure 12
1.3.2 0D and 2D hybrid Optoelectronic Application 12
1.4 References 15
2. High Electrical Performance Large-Scale Molybdenum Disulfide Field-Effect Transistor Fabricated with Various Dielectric Materials as the Back Gate 26
2.1 Introduction 26
2.1.1 Transition Metal Dichalcogenides (TMDCs) 26
2.1.2 Synthesis of TMDCs 28
2.1.3 Mechanism of CVD Process. 30
2.1.4 Type of CVD process 31
2.1.5 MOCVD process for large scale TMDCs 33
2.2 Experiment section 34
2.2.1 Materials 34
2.2.2 Growth of few-layer MoS2 35
2.2.3 Device Fabrication 35
2.2.4 Characterization and electrical measurement method 36
2.3 Results and discussion 37
2.3.1 Synthesis of large scale MoS2 37
2.3.2 Characteristics of large scale MoS2 38
2.3.3 Fabrication method 45
2.3.4 MoS2 FETs with Various Dielectric Materials as the Back Gate 48
2.4 Conclusion 51
2.5 References 52
3. Tunable Wavelength, Charge Carrier Transfer of Hybrid 0D-2D Lead halide Perovskite Quantum Dot-Molybdenum Disulfide Photodetectors 62
3.1 Introduction 62
3.2 Experiment section 64
3.2.1 Materials 64
3.2.2 Synthesis of PQDs 64
3.2.3 Bilayer MoS2 growth 66
3.2.4 Device fabrication 66
3.2.5 Characterization of optical and microstructure 67
3.2.6 Electrical and photocurrent measurement 69
3.3 Results and discussion 69
3.3.1 Characteristics of hybrid structure with PQDs and MoS2 69
3.3.2 Band alignment and charge transport of hybrid structure 72
3.3.3 Wavelength dependence of hybrid photodetector 76
3.3.4 Hybrid photodetector on 638 laser 78
3.4 Conclusion 85
3.5 References 86
4. Ag2Te/MoS2 hybrid photodetector with high detectivity and fast response in the infrared region 93
4.1 Introduction 93
4.2 Experiment section 95
4.2.1 Materials 95
4.2.2 Synthesis of Ag2Te QDs 95
4.2.3 Growth of few-layer MoS2 96
4.2.4 Device Fabrication 97
4.2.5 Characterization method for hybrid structure 97
4.2.6 Electrical and photocurrent measurement 98
4.3 Results and discussion 98
4.3.1 Characteristics of Ag2Te Quantum dots 98
4.3.2 Hybrid Structure of Ag2Te QDs and MoS2 101
4.3.3 Ag2Te/MoS2 hybrid IR photodetector 104
4.3.4 Response time of Ag2Te/MoS2 hybrid photodetector 108
4.4 Conclusion 111
4.5 References112
5. 요약문 117DoctordCollectio
비틀린 그래핀 다중층의 양자 홀 효과
No full textTwisted multi-layer graphene, Interlayer coherence, interlayer tunneling, interlayer interaction, Bose-Einstein condensation, Exciton condensation, Quantum Hall effect|비틀린 그래핀 다중층, 층간 결맞음. 층간 터널링, 층간 상호작용력, 보즈-아인슈타인 응축, 엑시톤 응축, 양자 홀 효과Both interlayer interaction and interlayer tunneling is key parameter in multilayer system. Grahene, with its spin-valley symmetry and adherence to the Dirac equation, serves as a platform for exploring novel quantum states not observed in conventional two-dimensional electron system. While twisted bilayer and trilayer graphene with small twist angles, where band hybridization occurs, are actively studied, twisted system with large angles remain underexplored. Here we explore various quantum states arising from strong interlayer interactions in twisted multilayer graphene systems. First, we made twisted bilayer graphene with a large twist angle to suppress interlayer tunneling through momentum mismatch, allowing the layers to remain decoupled without band hybridization. Despite this decoupling, the system exhibits ultra-strong interlayer interactions, enabling the observation of interlayer-coherent states, specifically exciton condensation-driven odd integer quantum Hall states, in both the lowest and second Landau levels that are not observed in conventional GaAs- based double quantum well structures. At higher magnetic fields, we further observed a range of quantum Hall states and phase transitions in the lowest Landau level as a function of the displacement electric field. Notably, we identified the 1/3 quantum Hall state, confirmed through Monte Carlo calculations as the Halperin 333 state, which had been theoretically predicted but never observed in GaAs double quantum wells. This odd integer quantum Hall state was also observed in twisted double bilayer graphene, though interestingly, it only appeared when the orbital index of the bilayer graphene’s Landau level was 1, and was absent when the orbital index was 0. Expanding our study to twisted trilayer graphene, we examined quantum Hall states in configurations with both unidirectional and bidirectional twists. Distinct quantum Hall states emerged depending on the twist direction, and unlike in twisted bilayer graphene, we observed asymmetry between electron-side and hole-side quantum Hall effects and their phase transitions. Through magnetic field-dependent measurements, we found that various quantum Hall states originate from different sources. The findings underscore the need for further theoretical calculations to accurately determine the ground states of the quantum Hall effects in twisted trilayer graphene. These results highlight that large-angle twisted graphene systems, beyond small-angle hybridization regimes, host diverse and complex quantum phenomena driven by strong interlayer interactions, expanding the landscape of quantum states accessible in graphene-based heterostructures.|본 논문은 고유한 층간 비틀림 각도를 갖는 비틀린 다층 그래핀 시스템에서 관찰된 양자 홀 효과에 중점을 둔다. 먼저, 큰 비틀림 각도를 가진 비틀린 이층 그래핀을 층간 응집 엑시톤 응축체를 탐구하기 위한 이상적인 플랫폼이자 초강력 층간 상호작용을 나타내는 새로운 2차원 시스템으로 소개한다. 기존의 GaAs 이중 양자 우물 구조에서는 채움계수가 1일 때 층간 결맞음에 의해 Halperin (111) 파동함수로 설명되는 양자 홀 상태가 관찰되었지만, 채움계수가 3과 1/3 일 때 층간 결맞음 상태는 이론적으로 예측되었음에도 불구하고 GaAs 시스템에서는 관찰되지 않았다. GaAs 이중 양자 우물 구조는 일반적으로 터널링을 방지하기 위해 20 nm의 장벽이 필요하여 층간 거리(d)가 자기 길이(lB)보다 큰 영역으로 탐구가 제한되고, 이에 따라 게이팅의 유연성도 제한된다. 이와 대조적으로, 비틀린 이층 그래핀은 고유한 이점을 제공한다. 비틀린 이층 그래핀은 터널링 장벽이 없고 층간 거리가 0.34 nm에 불과해 강력한 층간 상호작용을 허용하며, 큰 비틀림 각도로 얻은 모멘텀 불일치를 통해 층간 터널링을 억제할 수 있다. 또한 게이팅이 용이하여 더 넓은 매개변수 공간을 탐구할 수 있는 장점이 있다.
이러한 이점을 통해 d/lB ≈ O(10−2)<<1 영역에 도달할 수 있었고 두 번째 란다우 준위에서 층간 결맞음을 가진 홀수 정수 양자 홀 상태를 관찰할 수 있었다. 또한 동일한 비틀린 이층 그래핀 시스템으로 더 높은 자기장에서 상부 및 하부 게이트를 통해 밀도와 변위 전기장을 조절하여 최하위 란다우 준위에서 여러 분수 양자 홀 상태와 그들 간의 상전이를 관찰하였다. 특히 1/3 채움 계수에서 분수양자홀 효과를 발견하였고 Monte-Carlo 계산을 통해 Halperin 333 상태에 해당함을 확인하였다.
비슷한 홀수정수 양자홀 효과를 비틀린 이중 이층 그래핀에서도 관찰하였으며, 이 상태는 부분적으로 채워진 란다우 준위의 오비탈 지수가 1을 가질 때에만 나타난다는 점을 발견하였다. 이는 오비탈 지수가 1일 때 스핀 텍스처에 의한 여기가 엑시톤 상태에 주요한 영향을 미친다는 연구결과와 일치한다. 마찬가지로 밀도와 변위 전기장을 조절하여 전체 채움 계수가 -3/2 일 때 나타나는 분수 양자 홀 효과도 발견하였으며, 이는 오비탈 지수가 0인 이층 그래핀에서 오는 −1 정수 양자 홀 상태와 오비탈 지수가 1인 이층 그래핀에서 p-wave 페어링된 복합 페르미온에 의해 나타나는 −1/2 분수 양자 홀 상태의 조합으로 발생하는 것으로 예상된다.
더 나아가, 5도의 각도로 단방향 및 양방향으로 회전된 두 종류의 비틀린 삼중층을 탐구하였다. 측정 결과, 비틀림 방향에 따라 고유한 양자 홀 상태가 나타나는 것을 확인하였다. 삼중층 그래핀이 ABC 및 ABA 구조에 따라 다른 밴드를 나타내듯이, 비틀린 삼중층 그래핀도 비틀림 방향에 따라 다른 층간 전자 이동 변수로 인해 상이한 양자 홀 상태를 나타낼 것으로 예상된다.
이러한 발견은 비틀린 다층 그래핀 시스템의 전자적 특성에 층간 상호작용과 터널링이 미치는 중요한 영향을 보여준다. 본 연구는 현재 활발히 연구 중인 작은 각도에서의 밴드 혼성화뿐만 아니라 큰 비틀림 각도에서도 다양한 양자 상태가 존재함을 강조하며, 그래핀 기반 이종 구조에서 접근 가능한 양자 현상의 범위를 넓히는 데 기여한다.Abstract i
List of contents ii
List of figures iv
List of tables xi
Ⅰ. Introduction 1
1.1 Quantum Hall effect in single layer system 2
1.1.1 Landau level 2
1.1.2 Integer, fractional quantum Hall effect 5
1.1.3 Quantum Hall effect in graphene, bilayer graphene 11
1.2 Quantum Hall effect in double layer system 13
1.2.1 Quantum Hall effect at ntot = 1 13
1.2.2 Exciton condensation in double layer system 15
1.3 Twisted bilayer graphene 24
1.3.1 Small angle twisted bilayer graphene 24
1.2.2 Large angle twisted bilayer graphene 25
ⅠⅠ. Integer quantum Hall effect in large angle twisted bilayer graphene 27
2.1 Structure of large-angle twisted bilayer graphene 27
2.2 Magneto transport data under the low perpendicular magnetic field 29
2.3 Magneto transport data under the high perpendicular magnetic field 31
2.4 Theoretical model 38
2.4.1 The Hartree-Fock method 39
2.4.2 Mean field solutions 40
2.5 Ground states of quantum Hall effects in N = 1LL 45
2.6 Energy gap of integer quantum Hall states 48
2.7 Odd-integer quantum Hall states in N = 0LL 50
2.8 Discussion between low and large twist angle devices 53
2.9 In-plane magnetic field measurements 55
2.10 Conclusion 57
IⅠⅠ. Fractional quantum Hall effect in large angle twisted bilayer graphene 58
3.1 Sample structure and images 58
3.1 Magneto transport data under the high perpendicular magnetic field 60
3.2.1 Main data of device D1 60
3.2.2 Additional transport data of device D2, D3, and D4 64
3.2.2 Interlayer capacitance model 70
3.3 Theoretical Model 74
3.3.1 Model Hamiltonian 74
3.3.2 Candidate wave functions 76
3.3.3 Energy calculation method 78
3.2.4 Multi-component states 81
3.4 Theoretical Analysis 83
3.4.1 Main results: Phase diagram 83
3.4.2 Summary of Monte Carlo energy simulations for various FQH-states 90
3.5 Conclusion 102
ⅠV. Quantum Hall effect in large angle twisted double bilayer graphene 103
4.1 Gap opening in twisted double bilayer graphene 103
4.2 Integer quantum Hall states in twisted double bilayer graphene 106
4.3 Fractional quantum Hall states in twisted double bilayer graphene 111
4.4 Conclusion 114
V. Quantum Hall effect twisted trilayer graphene 115
5.1 Sample structure and fabrication 115
5.1 Twisted trilayer graphene with chiral rotation 117
5.2 Twisted trilayer graphene with alternative rotation 130
VⅠ. Summary 136
VⅠI. Appendix 138
7.1 Device fabrication 138
7.1.1 Dry pick up method 138
7.1.2 Full-dry flipping transfer method 138
7.1.3 Electrode-free AFM anodic oxidation lithography 141
7.1.2 E-beam lithography, RIE, and metal evaporation 142
7.2 Transport measurement 142
References 143DoctordCollectio
예쁜꼬마선충의 신경계와 근육계의 모델 개발을 통한 커넥톰, 신경 활성 및 행동 재현
No full textC. elegans, Neuronal network modeling, Newtonian mechanics, Simulation, Combined error function, 예쁜꼬마선충, 신경망 모델링, 뉴턴역학모델, 시뮬레이션, 결합형 에러 함수I. Introduction 1
1.1 Caenorhabditis elegans 1
1.2 Neuronal network modeling of C. elegans 1
1.3 Structural data of C. elegans 2
1.4 Functional data of C. elegans 2
1.5 Challenges in integrating structural and functional data 2
1.6 Thesis contributions 3
II. ElegansBot: Development of equation of motion deciphering locomotion including omega turns of Caenorhabditis elegans 5
2.1 Motivation 5
2.2 Results 6
2.2.1 Newton's equation of motion for locomotion of Caenorhabditis elegans: How does ElegansBot work? 6
2.2.2 Can C. elegans in ElegansBot crawl or swim? 10
2.2.3 ElegansBot exhibits more complex behavior including the turn motion. 15
2.2.4 ElegansBot presents body shape ensembles of C. elegans from a shape in water en route to agar. 18
2.3 Discussion 23
2.3.1 ElegansBot is an advanced kinetic simulator that reproduces C. elegans' various locomotion. 23
2.3.2 ElegansBot will serve as a strong bridge for enhancing the knowledge in "from-synapse-to-behavior" research. 23
2.4 Methods 24
2.4.1 Frequency and wavelength of C. elegans locomotion 24
2.4.2 C. elegans locomotion videos 24
2.4.3 Obtaining kymograms from video 24
2.4.4 Program code and programming libraries 25
2.4.5 Physical constants of the ground surface 25
2.4.6 Defining behavioral categories 25
2.4.7 Worm's mass, actuator elasticity coefficient, and damping coefficient 27
2.4.8 Minimum information required to describe the motion of each rod 28
2.4.9 Preservation of linearity in friction 29
2.4.10 Frictional torque by rotational motion 30
2.4.11 Proof of muscle force 31
2.4.12 Joint force calculation method 32
2.4.13 Proof of numerical integration for the translational motion of a worm using semi-implicit Euler method 36
2.4.14 Numerical integration of the rotational motion of i-rod using semi-implicit Euler method 38
2.4.15 Correction formula for the rotational inertia of the entire worm 40
2.4.16 Proper selection of friction coefficients 46
III. Reverse-engineering of functional connectome weights of Caenorhabditis elegans from its behaviors in experiments 48
3.1 Motivation 48
3.2 Background 50
3.3 Results 52
3.3.1 Functional weights of gap junctions and chemical synapses were reversed-engineered with high correlations to anatomical connectomes. 52
3.3.2 Spatio-temporal track of locomotion and membrane potential of SMD neurons of C. elegans were reproduced using reverse-engineered functional weights of gap junctions, chemical synapses, and leaky-integrator equation. 54
3.3.3 The impact of trp-1,2 mutation and SMD(D/V) ablation observed in the proprioception experiments of C. elegans was explained and reproduced. 57
3.3.4 Central pattern generators for periodic locomotion of C. elegans were identified and their role for the locomotion was explained. 60
3.4 Discussion 62
3.4.1 CANN, the functional weights of connectome, explained and reproduced various experimental results of C. elegans. 62
3.4.2 Method evaluating the success in reverse-engineering the nervous system of C. elegans 62
3.4.3 Application of the new strategy to other nervous system 63
3.4.4 Limitations 63
3.5 Methods 65
3.5.1 Acquisition and modifications of anatomical connectome data 65
3.5.2 Inference of time constant of C. elegans neuron 65
3.5.3 Leaky-integrator equation for membrane potential calculation 66
3.5.4 Configuration of the membrane potential of command neurons 67
3.5.5 Numerical method for solution of leaky-integrator equation 68
3.5.6 Definition of the error function 68
3.5.7 Update process of functional weights for minimization of the error function 70
3.5.8 Stepwise neuron elimination 71
3.5.9 Critical neuron identification 72
3.5.10 Analysis tools 72
3.5.11 Induction of the leaky-integrator equation 72
3.5.12 Modified semi-implicit Euler method 73
3.5.13 Definition of target degree of muscle contraction 75
3.5.14 Numerical integration of leaky-integrator equation 76
3.5.15 Derivatives of the error function 77
3.5.16 Adaptive error minimization 80
3.5.17 Method to calculate body control angle 81
IV. Conclusion 82
References 84
국문요약 90DoctordCollectio
재료 엔지니어링을 통한 양자점 광전소자 특성제어
No full text양자점, 광전소자, 재료Colloidal quantum dots (CQDs) are promising nanomaterials for optoelectronic applications owing to their unique electronic properties and facile processability. The lead sulfide (PbS) CQDs are mainly used as photoactive materials in CQD optoelectronics due to their infrared (IR) absorption capability with easily tunable absorption range and high absorption coefficient. Realizing the efficient CQD optoelectronic devices demand the harmonious development of CQD surface chemistry and device architecture. In particular, the development of CQD surface passivation strategies based on molecular lead halide (PbX2, X=I, Br) and 1,2 ethanedithiol (EDT) provided a foundation to achieve the recent air-stable and efficient CQD optoelectronoic devices. However, despite these recent advancements, there are several challenges remaining in CQD optoelectronics hinder further improvements. In the CQD photovoltaics, the Fermi level (EF) mismatch between the materials used devices causes undesirable energy band alignment for charge extraction. Additionally, the surface cracks of hole transporting CQD layer results from solid-state ligand exchange process induces interfacial traps by permitting the metal electrode penetration. Also, in the short wave infrared (SWIR) photodetector applications, the non-polar surface facet characteristics of CQDs which band gaps are within SWIR region makes them vulnerable from oxidation and aggregations resulting low device performances. In this study, these issues are addressed by diverse material engineerings. The band alignment and interface issues of CQD photovoltaics are tailored by developing the polycatechol functionalized MXene (PCA-MXene). The functionalization of MXene surfaces with polycatechol enables the homogeneous dispersion of MXene in diverse organic solvents that used in the fabrication of CQD photovoltaic process, resulting the effective combination of MXene and CQDs. This inducing the modification of work function of CQDs through the charge redistribution by surface dipole of MXene, ultimately promoting hole extraction in CQD photovoltaics. Moreover, the surface crack issue is addressed by the 2D nano structure of MXene. The PCA-MXene employed as an interlayer inhibits metal electrode penetration into photoactive layer by covers the surface cracks present in the hole transporting CQD layer. Owing to these advantages, the CQD photovoltaics incorporated with PCA-MXene achieve a power conversion efficiency (PCE) of 13.6%, accompanied by enhanced thermal stability, compared to the control device with a PCE of 12.8%. Also, by developing the proper SWIR CQD purification and surface manipulation processes by adjusting the alkali metal and halide ligands, the CQD photodetector external quantum efficiency (EQE) of 80% and responsivity of 1.0 (A/W) at 1550 nm wavelength is achieved. . Keywords: quantum dot, optoelectronic, Work function, interface, surface chemistry 유 형 렬. Hyung Ryul You. Tailoring the electrical-physical properties of colloidal quantum dot optoelectronics via material engineering. Department of Energy Science and Engineering. 2025. 47p. Advisors Prof. Jongmin Choi. Co-Advisors Dae-Hyun Nam. Ph.D./ES 201924011 List of Contents
|양자점은 물질의 화학 조성 변화 없이 결정 크기변화를 통해 전자구조를 제어할 수 는 특징을 가지고, 저비용 공정이 가능한 물질로, 차세대 나노소재로 유망한 물질이다. 그중 황화납 (PbS) 양자점은 근적외선 및 단파 적외선 영역을 타겟으로 밴드갭 조절이 용이하며, 우수한 흡광 특성을 가져 다양한 양자점 광전소자로 응용 가능하다. 양자점 광전소자는 근 십수년 간 급격한 발전을 이룸에도 불구하고 효율 상승을 억제하는 페르미레벨 불일치, 계면 크랙, 불안정한 표면화학 문제가 존재한다. 본 논문에서는 신소재인 폴리카테콜 화 맥신 물질 (PCA-MXene) 을 양자점 태양전지에 도입하였다. 양자점 박막에 균일하게 도핑된 맥신 소재의 쌍극자는 양자점의 일함수 조절을 통해 정공전달에 유리한 에너지밴드를 형성 하게 한다. 또한 MXene 소재의 2D 구조 도입을 통해 정공전달층 크랙 사이로 금속 전극이 침투하는 문제를 해결하였다. 해당 전략을 통해 양자점 태양전지의 효율을 기존 12.8% 대비 13.6%로 향상시켰고, 맥신 표면의 소수성인 폴리카테콜에 의해 소자내 수분침투를 억제하여 열안정성 또한 30% 수준 향상시켰다. 또한 1550 nm 적외선 양자점 기반 포토디텍터 구현을 위해 양자점 분야에서 문제로 대두되던 표면화학 문제를 해결했다. 완전 불활성 기체 공정 도입을 통한 양자점 정제과정 제어 및 기존대비 증가된 Br 기반 리간드를 통해 1550 nm 파장대에서 응답도 1.0 A/W 및 양자효율 80%을 가지는 양자점 포토디텍터를 개발하였다.List of Contents
Abstract 1
List of contents 2
Ⅰ. INTRODUCTION
1.1 Lead sulfide CQDs for optoelectronics 3
1.2 Conventional issues in CQD optoelectronics 4
1.3 MXene engineering for CQD photovoltaics 5
1.4 Infrared CQD enineering for photodetectors 6
Ⅱ. RESULTS AND DISCUSSION
2.1.1 Conventional CQD photovoltaics and discussions 7
2.1.2 Organic solvent dispersible MXene for CQD photovoltaics 10
2.1.3 MXene doping tailors the work function of CQDs 16
2.1.4 Interface engineering of CQD photovoltaics 24
2.1.5 MXene integrated CQD photovoltiaic characterizations 29
2.2.1 Synthesis of CQDs for SWIR photodetectors 33
2.2.2 Surface chemistry of SWIR CQDs for photodetectors 34
2.2.3 Architecture of CQD photodetectors and characterizations 35
Ⅲ. CONCLUSION
3.1 Conclusion and future perspective of CQD optoelectronics .. 39
Ⅳ. EXPERIMENTAL
. 40
Ⅴ. REFERENCES
. 44DoctordCollectio
유전학 및 후성유전학 변화를 통한 이즈힙 단백질의 조절
No full textEZHIP, Metformin, H3K27me3, KDM6, Cancer EpigeneticsAbstract
List of Contents
Abbreviations
List of Tables
List of Figures
Ⅰ. Introduction
1. EZHIP overexpression inhibits PRC2 in various cancers
2. PFA Ependymoma is a fatal pediatric brain tumor
3. Metformin treatment reverses EZHIP expression and restores H3K27me3 levels
4. KDM6 is the potential target of metformin treatment
ⅠⅠ. Results
1. Effect of metformin on EZHIP-expressing cells
1.1. EZHIP levels are increased in U2OS
1.2. Metformin is cytotoxic to EZHIP-expressing cell lines
1.3. Metformin induces EZHIP reduction and H3K27me3 upregulation in EZHIP-expressing cells
1.4. Metformin alters EZHIP expression at the transcriptional level
2. Regulation of EZHIP by DNA Methylation
2.1. DNA demethylation induces EZHIP expression
2.2. Endogenous EZHIP expression is determined by DNA methylation
2.3. Metformin leads to subtle methylation in the EZHIP gene body
3. Regulation of EZHIP via KDM6 Family Reduction by Metformin
3.1. Metformin interacts with the histone demethylase KDM6 family
3.2. Metformin downregulates KDM6 family
3.3. KDM6A and KDM6B are potential targets for EZHIP reduction
ⅠⅠⅠ. Discussion
ⅠV. Conclusion and Future Perspective
V. Method and Materials
1. Cell Culture
2. Cell Viability Assay
3. Histone Extraction
4. Western Blotting
5. Quantitative RT-PCR (RT-qPCR)
6. Bisulfite sequencing
7. Chromatin Immunoprecipitation RT-qPCR (ChIP-qPCR)
8. Transformation and Plasmid Preparation
9. Stable Cell Line Generation
10. siRNA Transfection
11. Quantification and Statistical Analysis
References
Summary (Korean)MasterdCollectio
SN2-mediated decoupled precursor provision enables large-scale production of monodisperse lead halide perovskite quantum dots in a single reactor
No full textQuantum-confined lead-halide perovskite nanocrystals (QPNCs) are a promising optoelectronic semiconductor owing to their exceptional fluorescence and the size- and dimension-tunable optical properties. QPNCs having low formation energy encounter challenges in accurately regulating the nucleation and crystal growth stages during injection-based syntheses using lead halide reagents. Here, we introduce a non-injection, one-pot synthetic approach based on bimolecular nucleophilic substitution (SN2) and thermolysis reactions of the decoupled metal and halide precursors for the large-scale production of monodisperse CsPbX3-QPNCs (X = Cl, Br, I). This approach facilitates a homogeneous supply of halide anions and metal cations, enabling the precise control over the nucleation and crystal growth stages in the isolated size-focused region. Monodisperse CsPbX3-QPNCs achieve high color purity across the RGB color gamut by adjusting size, dimensionality, and halide composition, and can be produced on an ultra-large scale.FALSEsciescopu
BCL6 coordinates muscle mass homeostasis with nutritional states
No full textNutritional status is a determining factor for growth during development and homeostatic maintenance in adulthood. In the context of muscle, growth hormone (GH) coordinates growth with nutritional status; however, the detailed mechanisms remain to be fully elucidated. Here, we show that the transcriptional repressor B cell lymphoma 6 (BCL6) maintains muscle mass by sustaining GH action. Muscle-specific genetic deletion of BCL6 at either perinatal or adult stages profoundly reduces muscle mass and compromises muscle strength. Conversely, muscle-directed viral overexpression of BCL6 significantly reverses the loss of muscle mass and strength. Mechanistically, we show that BCL6 transcriptionally represses the suppressor of cytokine signaling 2 to sustain the anabolic actions of GH in muscle. Additionally, we find that GH itself transcriptionally inhibits BCL6 through the Janus kinase and signal transducer and activator of transcription 5 (JAK/STAT5) pathway. Supporting the physiologic relevance of this feedback regulation, we show the coordinated suppression of muscle Bcl6 expression with the induction of GH in the fasted state. These findings reveal the complexity of the feedback controls modulating GH signaling and identify BCL6 as a key homeostatic regulator coordinating muscle mass with nutrient availability. Moreover, these studies open avenues for targeted therapeutic strategies to combat muscle-wasting conditions. Copyright © 2025 the Author(s).FALSEsciescopu