Daegu Gyeongbuk Institute of Science and Technology
DGIST Library Institutional RepositoryNot a member yet
12664 research outputs found
Sort by
InAs 나노와이어의 조정 가능한 다중 양자점: Su-Schrieffer-Heeger 위상적 사슬을 위한 기초 구성 요소
No full textInAs nanowire, quantum dotList of Contents
Abstract i
List of contents ii
List of figures V
Ⅰ. Theoretical Background
1.1 Su-Schrieffer-Heeger (SSH) model 3
1.2 Quantum dots 10
1.2.1 Indium Arsenide (InAs) nanowire single-electron transistor 10
1.2.2 Single quantum dot 11
1.2.3 Double quantum dot 14
1.2.4 Quadruple quantum dot 19
1.3 Superconducting coplanar waveguide (SCPW) resonators 21
1.3.1 Microwave theory of transmission lines 22
1.3.2 Superconducting microwave resonators 24
1.3.3 Norton equivalent circuit 26
1.3.4 Quality factors of the resonator 28
1.3.5 Transmission coefficient 30
II. Fabrication
2.1 InAs nanowire single-electron transistor 32
2.1.1 Nanowire transfer and etching process 32
2.1.2 DC Device Fabrication 34
2.2 Nb Superconducting coplanar waveguide (SCPW) resonator 35
2.2.1 Coplanar waveguide resonator geometry 35
2.2.2 RF Device Fabrication 36
III. Measurement Techniques
3.1 Initial Characterization and Sample Preparation 39
3.2 Measurement Schematic 41
3.3 Low Temperature Measurement Setup 42
3.4 He-3 Refrigeration System 44
IV. Experiments Results
4.1 Multi-quantum dot (DC Device) Results 46
4.1.1 Single quantum dots 46
4.1.2 Double quantum dots 49
4.1.3 Triple quantum dot 51
4.1.4 Quadruple quantum dot 53
4.2 SCPW resonator (RF Device) Results 59
4.2.1 Characteristics of the RF Device 59
4.2.2 Interaction between the transistor and the resonator 62
V. Summary and Conclusion 64
Bibliography 65
Appendix
Fabrication Details and Recipes
A1. Wafer Characteristics of DC device
A2. Photo pad Fabrication
A3. Wafer Characteristics of RF device
A4. Resonator fabrication for measurements
A5. InAs transistor fabrication (DC & RF device)
A6. Designing using K-Layout and Layout Editor
Pre-Cooldown Sample Preparation
B1. Wire bonding
B2. He-3 Sample Exchange Procedure (Unloading and loading Sample)
Korean Summary 80MasterdCollectio
Investigating the role of PTP1B in neurodegenerative mechanisms induced by ALS/FTD-associated RNA-binding proteins
No full textNeurodegenerative diseases, RBP, Neuroinflammation, Protein quality contorl, PTP1BChapter Ⅰ. Introduction 1
1.1 Research background and purpose 1
1.2 Neurodegenerative disease (ALS and FTD) 1
1.3 RNA binding protein associated with ALS and FTD 4
1.4 Neuroinflammation 8
1.5 Cellular processes of protein quality control system 8
Chapter Ⅱ. Materials & Methods 11
2.1 Reagents and Antibodies 11
2.2 Primary cell cultures 11
2.3 NSC-34 cell culture 12
2.4 Cell line 12
2.5 Transfection 13
2.6 Cytotoxicity test 13
2.7 CMFDA staining analysis 14
2.8 Astrocyte-conditioned media (ACM) 14
2.9 Cortical neuron-astrocytes co-culture assay 15
2.10 ACM-treated neuron culture 15
2.11 Quantitative RT-PCR 15
2.12 Immunoblot analysis 16
2.13 Nuclear and cytoplasmic fraction extraction 16
2.14 Preparation of soluble and insoluble cell extracts 16
2.15 ELISA 17
2.16 Immunocytochemistry analysis 17
2.17 Mitochondrial activity assay 17
2.18 Quantification of dendritic spines 18
2.19 Human primary fibroblasts 18
2.20 Flow Cytometry Analysis 19
2.21 CYTO-ID® Autophagy Analysis 19
2.22 Drosophila Immunohistochemical analysis 19
2.23 Fly stains 20
2.24 Lifespan and climbing assays 20
2.25 Statistics 20
Chapter Ⅲ. Results 21
Part 1. Non-cell autonomous effects of astrocytes in contributing to neuronal cell death 21
Part 1-1. The role of PTP1B inhibition in alleviating neuroinflammation associated with TDP-43 22
3.1 Investigating the role of PTP1B in CNS 22
3.2 PTP1B is an essential modulator of TDP-43-induced inflammation in astrocytes 24
3.3 PTP1B regulates TDP-43-induced inflammation via the NF-κB pathway 32
3.4 PTP1B inhibition and absorption of proinflammatory cytokines mitigate the neuronal toxicity caused by astrocytic TDP-43 overexpression 37
3.5 PTP1B inhibition suppresses astrocytic TDP-43-induced mitochondrial dysfunction and spine retraction in neurons 45
3.6 PTP1B inhibition ameliorates astrocytic TDP-43-induced neuronal toxicity and mitochondrial dysfunction in motor neuron-like cells 50
3.7 Inflammation and neuronal toxicity induced by glial TDP-43 are mitigated by PTP1B downregulation in Drosophila 55
Part 1-2. The effects of DHE in mitigating neuroinflammation related to mutant FUS in astrocytes 65
3.8 Investigating the role of natural compound-derived PTP1B inhibitors in CNS 65
3.9 DHE alleviates aggregation of FUS P525L and activation of inflammatory response in mutant FUS P525L overexpressed astrocytes 66
3.10 DHE treatment reduces astrocytic FUS P525L-induced neuronal toxicity and mitochondria dysfunction 73
3.11 DHE treatment rescues glial FUS induced toxicity in Drosophila 76
3.12 DHE treatment moderates inflammatory response in FUSALS patient fibroblast 83
Part 2. Cell-autonomous effects of TDP-43 on UPS impairment 88
3.13 The mitigating effects of PTP1B inhibition on UPS impairment dysregulation in neuronal cells 89
3.14 Induced ALP by TDP-43 overexpression is restored though inhibition of PTP1B in N2a 96
3.15 Inhibition of PTP1B ameliorates TDP-43-induced neurotoxicity in vitro and in vivo 101
3.16 PTP1B overexpression exacerbates TDP-43-induced neurotoxicity 106
Chapter Ⅳ. Discussion 113
References 120
Abstracts in Koreans 133DoctordCollectio
애기장대에서 일주기 시계와 노화의 기능적 관계
No full textaging, circadian clocks, gene regulation, diurnal rhythmsOne of the most remarkable biological systems evolved by life on Earth is programmed senescence, a mechanism that facilitates generational turnover and drives evolutionary progress. Given the intricate and high- stakes molecular processes involved, organisms utilizing this system must precisely monitor their aging. This raises fundamental questions: how do organisms perceive the passage of time, and what systems enable this capability? Aging research has identified specific regulator genes that either promote or inhibit aging, demonstrating that manipulating these genes can modulate the aging rate. Here, I propose the existence of a biological oscillator that enhances the precision of aging, complementing established cumulative aging mechanisms. Using Arabidopsis thaliana as a multicellular model system due to its genetic, genomic, and developmental tractability, I identified PRR9, a core circadian clock component, and ORE1, a positive regulator of leaf senescence, as critical factors in maintaining cell-to-cell synchronization associated with leaf senescence, thereby linking the circadian clock to the aging process at the functional level. Furthermore, my research revealed that age-related alterations in circadian clock activities precede other aging symptoms and manifest across organismal, organ, and cellular levels. Under diurnal conditions, these clock modifications affect time perception, reshaping diurnal gene expression patterns. Aged plants perceive a “shorter day and longer night” due to shifts in specific gene groups, including those involved in light responses. Importantly, these alterations are genetically programmed, as loss-of-function mutations in circadian clock genes, including CCA1, LHY, and PRR9, disrupt aging-dependent changes in the rhythmic gene expression. This thesis provides new insights into the role of biological oscillators in time perception and underscores their critical involvement in aging-related processes.|지구 생명체가 가진 가장 강력한 시스템 중 하나는 노화 프로그램이다. 노화는 세대 교체를 촉진하여 진화적 발전을 이끄는 메커니즘이다. 노화과정에 관여하는 분자적 장치들은 개체 스스로의 죽음을 유도하기 때문에, 이러한 시스템을 사용하는 생명체는 자신의 노화단계를 정밀하게 모니터링해야 한다. 이는 근본적인 질문을 제기한다. 생명체는 시간의 흐름을 어떻게 인지하는가? 노화 연구는 노화를 촉진하거나 억제하는 특정 유전자들을 밝혀내왔고, 이 유전자들을 조작함으로써 노화 속도를 조절할 수 있음을 보여주었다. 본 연구에서는 애기장대를 다세포 모델 시스템으로 활용하여 누적 노화 메커니즘을 보완하여 노화의 정밀성을 향상시키는 생물학적 진동자의 존재를 제안한다. 연구 결과, 일주기 시계의 핵심 구성 요소인 PRR9과 노화 조절 유전자 ORE1이 세포 간 동기화를 유지하는 데 중요한 역할을 하며, 이는 일주기 리듬과 노화를 연결하는 데 기여함을 확인했다. 나이에 따른 일주기 시계 활동의 변화는 다른 노화 증상보다 선행하여 나타나며, 유기체, 기관, 세포 수준에 걸쳐 관찰되었다. 주기적인 환경에서 이러한 시계 변화는 시간 인식에 영향을 미쳐, 일주기 유전자 발현 패턴을 재구성했다. 나이 든 식물은 특정 유전자 그룹, 특히 빛 반응에 관여하는 유전자들의 변화로 인해 “짧은 낮과 긴 밤”을 경험하게 된다. 중요한 점은 이러한 변화가 유전적으로 프로그램되어 있다는 것이다. CCA1, LHY, PRR9 등 일주기 시계 유전자에서 기능 상실 돌연변이가 발생하면, 노화에 따라 리듬 유전자 발현 변화가 방해받는 것으로 나타났다. 본 연구는 생물학적 진동자가 시간 인식에 미치는 역할에 대한 새로운 통찰을 제공하며, 이러한 진동자가 노화와 관련된 과정에서 중요한 역할을 한다는 점을 강조한다.List of Contents
Abstract i
List of Contents ii
List of Figures v
List of Table vii
I. Background Introduction
1.1 Organization of the Thesis 1
1.2 Senescence 3
1.2.1 The biological Role of Aging 3
1.2.2 Senescence in Arabidopsis 5
1.3 The Circadian Clock 8
1.3.1 Biological Functions of the Circadian Clock 8
1.3.2 The Circadian Clock in Arabidopsis 10
1.3.3 Mathematical Modeling for Circadian Clock Activities in Arabidopsis 13
1.3.4 Functional Links between the Circadian Clock and Aging 15
1.3.5 Cell-Type Specificity in the Arabidopsis Circadian Clock 17
1.4 Main Hypothesis and Specific Aims 19
II. Free-Running Circadian Clock Alterations as Early Indicators of Aging in
Arabidopsis: A Multi-Scale Investigation
2.1 Introduction 20
2.2 Materials and Methods 23
2.2.1 Plant Materials and Growth Conditions 23
2.2.2 Luciferase Reporter System 23
2.2.3 Detection of Cellular Clock Activity in Intact Plants 24
2.2.4 Photosynthetic activity Analysis 24
2.3 Results 26
2.3.1 Development of a Robust Luciferase Reporter System 27
2.3.2 Performance of the Upgraded Luciferase Reporter System 29
2.3.3 Circadian Period Length as a Marker for Senescence 31
2.3.4 PRR9 Maintains Cell-to-Cell Synchronization Linking the Circadian Period and
Aging 33
2.4 Discussion 36
2.5 Acknowledgement 37
III. The Genetically Programmed Rhythmic Alteration of Diurnal Gene Expression in
Aged Arabidopsis Leaves
3.1 Introduction 38
3.2 Materials and Methods 41
3.2.1 Plant Materials and Growth Conditions 41
3.2.2 Luciferase Assay 41
3.2.3 Physiological Time 41
3.2.4 Full Width at Half Maximum (FWHM) Calculation 42
3.2.5 Gene Expression Analyses 43
3.2.6 RNA sequencing (RNA-seq) Analysis 43
3.2.7 Selection of Genes with Oscillating Expression Patterns 44
3.2.8 Statistical Analysis 44
3.3 Results 46
3.3.1 Age-Dependent Rhythmic Alteration of Free-Running Circadian rhythms 46
3.3.2 Age-Dependent Alteration of Clock Gene Expression Under Diurnal conditions
49
3.3.3 Age-Dependent Alteration of Diurnal Transcriptomes 53
3.3.4 Senescence-Regulatory Gene Expression Patterns in Aged Leaves 54
3.3.5 Transcriptomic-Level Rhythmic Alteration of Daily Gene Expression Rhythms
55
3.3.6 Opposing Rhythmic Alterations in Daytime and Nighttime Genes in Aged Plants
59
3.3.7 Core Clock Components Mediate Rhythmic Alterations in Gene Expression 62
3.4 Discussion 65
3.5 Acknowledgement 69
IV. Appendix 70
4.1 Computational Simulation of Circadian Clock Dynamics in Arabidopsis 70
4.2 ORE1 Inhibits TOC1-Mediated PRR5 Expression 75
4.3 Acknowledgement 91
V. Conclusion 92
VI. References 93
VII. Abstract in Korean 103
VIII. Acknowledgement 104DoctordCollectio
형상 프로그래밍이 가능한 자기 소프트 로봇의 3D 프린팅
No full textSoft robot, Magnetic field, 3D printing, Photo-curing, Anti-settlingⅠ. INTRODUCTION 1
Ⅱ. BACKGROUND 5
2.1 Working mechanism of the magnetically actuated soft robots 5
2.2 LCD based photocuring 3D printing 7
Ⅲ. METHODS 10
3.1 Fabrication of the magnetically actuated soft robots 10
3.2 Actuation of the magnetically actuated soft robots 12
3.3 Measurement of the sedimentation of Magnetic particles 13
3.4 Magnetic response of the magnetic composite 13
3.5 Mechanical properties of magnetic composite 14
3.6 Exposure time of the magnetic composite 14
Ⅳ. RESULTS and DISCUSSION 16
4.1 Characterization of the magnetic composites 16
4.2 Characterization of MSLA 3D printer 24
4.3 Actuation of the Magnetically actuated soft robots 26
4.4 Biomedical application of the soft robot 31
Ⅴ. CONCLUSION 37
Ⅵ. REFERENCES 38MasterdCollectio
라이트닝 신경방사장을 위한 마이크로스케일링 포맷 양자화 최적화
No full textNeRF, QuantizationNeRF(Neural Radiance Fields)는 최근 신경 렌더링 분야에서 주목받고 있는 기술로, 2D 이미지 세트를 기반으로 복잡한 3D 장면을 새로운 시점에서 생성하는 데 사용된다. NeRF 는 명시적인 3D 모델을 구축하지 않고도 부드럽고 연속적인 시점 전환을 통해 고품질의 사실적인 렌더링을 제공하는 능력으로 많은 관심을 받고 있다. 기존의 3D 렌더링 방식이 메쉬, 포인트 클라우드, 복셀 등을 활용해 3D 장면을 렌더링하는 것과 달리, NeRF 는 신경망을 사용해 장면의 부피 밀도와 색상을 예측하고 이를 통해 3D 구조를 암시적으로 학습한다. 이러한 NeRF는 가상현실(VR), 증강현실(AR), 자율주행 등 다양한 분야에 응용되고 있다. 그러나 NeRF 는 특히 실시간 응용에서 여러 계산상의 한계에 직면하고 있다. Mildenhall 등이 제안한 초기 NeRF 모델은 고성능 GPU(NVIDIA V100)에서도 초당 0.03 프레임의 낮은 렌더링 속도를 보이며, 이는 상호작용이 중요한 VR 환경이나 자율주행과 같은 시간 민감형 응용에는 적합하지 않다. 따라서 최근 연구는 NeRF 의 효율성을 높이기 위해 렌더링 기술을 최적화하고, 이를 위한 전용 하드웨어 솔루션을 탐구하고 있다. 이 과정에서 양자화(Quantization), 특히 학습 후 양자화(PTQ, Post-Training Quantization)가 NeRF 모델의 계산 효율성을 높이는 유망한 방법으로 떠올랐다. PTQ 는 신경망의 가중치와 활성화 값을 고정소수점 형식으로 변환해 메모리 사용량과 계산 복잡도를 줄인다. 이를 통해 NeRF 는 더 적은 하드웨어 자원으로도 높은 품질의 렌더링을 수행할 수 있다. 또한 MX(Microscaling) 형식과 같은 새로운 데이터 형식은 더욱 효율적인 데이터 표현을 가능하게 하여 계산 비용을 줄일 수 있다. MX 형식은 블록 단위로 스케일링을 적용하여 부동소수점 및 정수 데이터를 효율적으로 표현하고, 메모리 사용량과 계산 오버헤드를 줄이는 데 기여한다. 본 연구에서는 이러한 양자화 기법과 MX 형식을 자율주행을 위한 NeRF 모델인 LightningNeRF 에 적용하여, 모델 크기와 추론 시간을 줄이는 동시에 렌더링 품질을 유지하는 방법을 탐구한다. 실험 결과, 이러한 양자화 전략이 NeRF 의 계산 복잡성을 효과적으로 줄이면서 실시간 응용에 더욱 적합한 성능을 발휘함을 보여준다.|NeRF (Neural Radiance Fields) is an emerging technology in the field of neural rendering that generates complex 3D scenes from sets of 2D images, allowing rendering from novel viewpoints. Unlike traditional 3D rendering methods that rely on meshes, point clouds, or voxels, NeRF leverages neural networks to predict volume density and color, implicitly learning the scene's 3D structure. Due to its ability to achieve high-quality, realistic rendering with smooth, continuous viewpoint transitions without explicit 3D models, NeRF has garnered significant interest. Applications span various domains, including virtual reality (VR), augmented reality (AR), and autonomous driving. Despite its advantages, NeRF presents several computational challenges, particularly in real-time contexts. The original NeRF model proposed by Mildenhall et al. achieves a low rendering speed of only 0.03 frames per second, even on high-performance GPUs like the NVIDIA V100, making it unsuitable for time-sensitive applications such as interactive VR environments or autonomous driving systems. Consequently, recent research has focused on optimizing NeRF rendering techniques and developing specialized hardware solutions to enhance its efficiency. Among these approaches, quantization—particularly post-training quantization (PTQ)—has emerged as a promising method to improve NeRF's computational performance. PTQ reduces memory usage and computational complexity by converting the neural network's weights and activations into a fixed-point format, enabling high-quality rendering with fewer hardware resources. Additionally, new data formats, such as the MX (Microscaling) format, facilitate more efficient data representation and reduce computational costs. The MX format applies block-wise scaling to effectively represent both floating-point and integer data, thereby contributing to reduced memory usage and computational overhead. This study investigates the application of quantization techniques, including the MX format, to Lightning-NeRF, a NeRF model optimized for autonomous driving. The goal is to minimize model size and inference time while maintaining rendering quality. Experimental results demonstrate that these quantization strategies significantly reduce the computational complexity of NeRF, making it more feasible for real-time applications. Keywords: NeRF, Quantization, Post-Training Quantization, MX Format, Lightning-NeRFList of Contents
Abstract i
List of contents ii
List of tables iii
List of figures vi
Ⅰ. Introduction 1
II. Background and Motivation
2.1 NeRF(Neural Radiance Fields) 3
2.2 PTQ(Post-Training Quantization) 4
2.3 MX(Microscailing) Format 5
2.4 Low Rendering Speed 7
III. Quantization Strategies
3.1 Target Model Setup 8
3.2 Target Model Analysis 8
3.3 Quantization Strategies for Lightning-NeRF 13
IV. Evaluation
4.1 Image Quality Results 24
4.2 Model Size 26
V. Conclusion and Future Work 27MasterdCollectio
새로운 리튬 고체 전해질의 발견: 신규 구조 유형과 높은 이온 전도도
No full textAll-solid-state batteries, solid electrolyte, Ionic conductivity, X-ray diffractionⅠ. Motivation
1.1 The need for developing all-solid-state-batteries 1
1.2 Why sulfide solid electrolytes? 2
1.3 Introduction to Li2GeS3 3
1.4 Introduction to Li21Ge8P3S34 5
Ⅱ. Experimental
2.1 Synthesis 7
2.1.1 Synthesis of Li2GeS3 7
2.1.2 Synthesis of Li21Ge8P3S34 7
2.2 Materials characterization 8
2.2.1 Structure analysis of Li2GeS3 8
2.2.2 Electrochemical analysis of Li2GeS3 10
2.2.3 Structure Analysis of Li21Ge8P3S34 11
2.2.4 Electrochemical analysis of Li21Ge8P3S34 14
Ⅲ. Results and discussions
3.1 Results of Li2GeS3 17
3.1.1 Structure Properties 17
3.1.2 Li+ ions Transport Properties 23
3.2 Results of Li21Ge8P3S34 29
3.2.1 Structure Properties 29
3.2.2 Electrochemical performance 41
Ⅳ. Conclusion 45
Ⅴ. References 46MasterdCollectio
Zn(ii)-driven impact of monomeric transthyretin on amyloid-β amyloidogenesis
No full textExtracellular accumulation of amyloid-β (Aβ) peptides in the brain plays a significant role in the development of Alzheimer's disease (AD). While the co-localization and interaction of proteins and metal ions with Aβ in extracellular milieu are established, their precise pathological associations remain unclear. Here we report the impact of Zn(ii) on the anti-amyloidogenic properties of monomeric transthyretin (M-TTR), which coexists spatially with Aβ and Zn(ii) in extracellular fluids. Our findings demonstrate the Zn(ii)-promoted ternary complex formation involving M-TTR, Aβ, and Zn(ii) as well as M-TTR's proteolytic activity towards Aβ. These interactions decrease the inhibitory effect of M-TTR on the primary nucleation process of Aβ as well as its ability to improve cell viability upon treatment of Aβ. This study unveils the variable activities of M-TTR towards Aβ, driven by Zn(ii), providing insights into how metal ions influence the entanglement of M-TTR in the Aβ-related pathology linked to AD. © 2025 The Royal Society of Chemistry.TRUEsciescopu
광학적 산화 제어 공정을 통한 금속 산화물 어레이 기반 차세대 인공 후각 시스템 개발
No full textArtificial olfactory system, Metal oxide semiconductor, Laser-induced oxidation, Oxygen vacancy control, Artificial neural network, Odorant detectionList of Contents
List of Contents
Abstract i
List of Contents iii
List of Figures vii
List of Tables xii
Ⅰ. Introduction
1.1 Overview of AOSs 1
1.1.1 Historical Development of AOSs 1
1.1.2 Principle of Existing AOSs 4
1.1.3 Applications and Limitations of Current Systems 6
1.2 Limitations and Challenges of Current AOSs 8
1.2.1 Challenges in Selectivity and Sensitivity 8
1.2.2 Challenges in Energy Efficiency and Scalability 10
1.2.3 Technological Barriers to Widespread Adoption 11
1.3 Next-Generation AOSs: Objectives and Innovations· 13
1.3.1 Advanced Sensor Technologies for Improved Performance 13
1.3.2 Integration of Signal Processing Techniques Using Machine Learning 20
1.3.3 Objectives for Next-Generation AOSs 26
1.4 Research Objectives and Scope 28
1.4.1 Olfactory Epithelium: Sensor Fabrication and Oxidation Process 29
1.4.2 Olfactory Bulb: Singal Generation and Odor Sensing Measurements 30
1.4.3 Olfactory Cortex: Signal Processing Using ANNs 32
1.4.4 Applications and Validation of the AOS through Wine Classification 35
ⅠI. Olfactory Epithelium: Sensor Fabrication and Oxidation Process
2.1 Overview of MOS Sensor Arrays 38
2.1.1 Principles of Olfactory Receptors in Biological Systems 41
2.1.2 Emulation of Olfactory Receptors Using MOS Sensors 43
2.1.3 Suitability and Advantages of MOS Materials for AOSs 45
2.1.4 Room-Temperature Operation of MOS Gas Sensors 47
2.2 Analysis of Material Properties through Laser Oxidation Process 52
2.2.1 Substrate Preparation and Deposition of Cu Thin Films 53
2.2.2 Laser Oxidation Process and Parametric Studies 54
2.2.3 Characterization of Oxidized Cu Films 58
2.3 Channel Diversification Strategy through Oxygen Vacancy Modulation via
Laser Oxidation Process 64
2.3.1 Design and Development of MOS Sensor 64
2.3.2 Fabrication Process Using Laser-Induced Oxidation 65
2.3.3 Characterization of Laser-Oxidized Films 68
2.3.4 Sensing Properties of Laser-Oxidized MOS Sensor 71
ⅠII. Olfactory Bulb: Signal Generation and Odor Sensing Measurements
3.1 Experimental Setup for Odor Sensing 74
3.1.1 Design of the Bubbler-based Odor Generation System 75
3.1.2 Sensor Array Configuration and Multi-Channel t-I System 77
3.1.3 Odor Generation and Control Protocols 78
3.2 Description of Selected Odor Molecules 80
3.2.1 Criteria for Odor Molecule Selection 80
3.3 Measurement of Single Odor Molecules 81
3.3.1 Sensor Response to Single Odor Molecules 81
3.3.2 Interpretation of Single Odor Response Patterns 90
3.4 Analysis of Complex Odor Mixtures 95
3.4.1 Sensor Response to Dual Odor Molecules Mixtures 95
3.4.2 Pattern Recognition for Complex Odor Molecule Mixtures 97
IV. Olfactory Cortex: Enhancing Efficiency with Neural and Preprocessing Techniques
4.1 Overview of Artificial and Biological Olfactory Systems 100
4.1.1 Applications and Energy Efficiency Challenges in AOSs 100
4.1.2 Biological Olfactory Systems as a Model for Energy-Efficient Signal
Processing 101
4.2 Signal Processing Techniques for Energy Efficiency 102
4.2.1 Sensor Array Design Aligned with Low Energy Efficiency Requirements · 102
4.2.2 Exponential Fitting by Langmuir Isotherm Model 104
4.2.3 Initial Signal Segment Analysis· 104
4.3 Odor Classification and Concentration Prediction Using SNN 106
4.3.1 SVM and SNN for Initial Signal Segment Classification 107
4.3.2 Evaluation of Classification Performance and Accuracy 108
4.3.3 Regression Techniques for Concentration Prediction 110
V. Applications and Validation of the AOS through Wine Odor Classification
5.1 Rationale for Extension from Odor Molecule Analysis 111
5.2 Wine Classification Results and Analysis 113
5.3 Broader Applications in Industry 117
5.3.1 Extending Olfactory Technology Beyond Wine Analysis 117
5.3.2 Implementation of Quality Control in Various Industries 118
5.3.3 Potential for Customizable Scent Profiles 118
5.3.4 Integration of AOSs in Real-World Scenarios 119
VI. Conclusion
6.1 Summary 120
6.2 Contributions and Future Directions 122
VII. References 123DoctordCollectio
MPS: Metric-aware Pseudo-label Selection Framework via Reinforcement Learning
No full textMachine LearningⅠ. Introduction
1 Introduction 1
ⅠI. Related work
2.1 Pseudo-label 3
2.2 Reinforcement learning 4
ⅠII. Method
3.1 Pre-trained teacher model optimize and pseudo labeling 8
3.2 Training data sampling by contribution 9
3.2.1 Task predictor optimization 9
3.2.2 Controller optimization 10
3.3 The reinforcement learning algorithm 10
3.4 Data contribution evaluation using reinforcement learning 14
IV. Experiments
4.1 Implement details 15
4.2 Result 17
4.2.1 SSL with selected data 17
4.2.2 Analysis 19
4.2.3 Discussion 20
V. Conclusion
5 Conclusion 22MasterdCollectio
