政府計畫（GRB），建議「依年度遞減排序」，以查看最新的研究方向。
| 畢業學年度 | 論文標題 | 連結 | 學位 | 畢業時長(years) |
|---|---|---|---|---|
| 關鍵字 | ||||
| 112 | 使用語音技... 使用語音技術量化研究自閉症光譜 (Using speech-based technology to characterize autistic traits of autism spectrum disorder) | NTHU | 碩逕博 | 無口試日期 |
自閉症(autism spectrum disorder)、語音訊號處理(speech signal processing)、深度學習(deep learning) 自閉症(a... | ||||
| 112 | 保護易受傷... 保護易受傷害人群隱私的自動語音辨識： 以阿茲海默症患者為例 (Privacy-Preserved Automatic Speech Recognition for Vulnerable Populations: A Case Study of People with Alzheimer's Disease) | NTHU | 碩 | 無口試日期 |
自動語音辨識(Automatic speech recognition)、阿茲海默症(Alzheimer’s disease)、隱私保護(privacy preservation)、聯邦學習(federated learning)、異質性(heterogeneity) 自動語音辨... | ||||
| 112 | MVRNe... MVRNet:基於單視角的著衣人體重建之多模態體積表示網絡 (MVRNet: Multimodal Volumetric Representation Network for Monocular Clothed Human Reconstruction) | NTHU NDLTD | 碩 | 2.23 |
單視角著衣人體重建(Monocular Clothed Human Reconstruction)、三維電腦視覺(3D Computer Vision)、多模態視覺(Multi-modal Vision)、參數化身體模型(Parametric body models) 單視角著衣... | ||||
| 112 | 基於循環式... 基於循環式對抗性網路的數據增量及梯度察覺純化防禦機制建立強健式語音情緒辨識模型 (Building a Robust Speech Emotion Recognition using Cycle-GAN-based Data Augmentation and Gradient-Aware Purification Adversarial Defense Mechanism) | NTHU NDLTD | 碩逕博 | 6.01 |
語音情緒辨識(Speech Emotion Recognition)、強健式學習(Robustness Learning)、循環式對抗性網路(Cycle Generative Adversarial Learning)、對抗式學習(Adversarial Learning) 語音情緒辨... | ||||
| 112 | 運用客製化... 運用客製化聯邦學習策略在日常環境中以生理訊號偵測壓力 (In-the-wild Physiological-based Stress Detection Using Customized Federated Strategy) | NTHU NDLTD | 碩 | 2.01 |
聯邦學習(Federated Learning)、聯邦遷移學習(Federated Transfer Learning)、深度互學習(Deep Mutual Learning)、知識蒸餾(Knowledge Distillation)、壓力偵測(Stress Detection)、心律變異分析(Heart Rate Variability) 聯邦學習(... | ||||
| 111 | 使用多提取... 使用多提取器在多說話人重疊問題中的個人語音活動檢測 (A Multi-Extractor Strategy for Personal-VAD in Multiple Speakers Overlapped Problem) | NTHU NDLTD | 碩 | 1.98 |
語者自動分段標記(speaker diarization)、語音活動檢測(PVAD)、多重提取器(multi-extractors)、重疊說話人(overlapped speech) 語者自動分... | ||||
| 111 | 轉換偵測解... 轉換偵測解碼器於語者預測 (Dialogical speaker decoder with transition detection for next speaker prediction) | NTHU NDLTD | 碩(提早入學) | 2.95 |
語者預測(next speaker prediction)、轉換偵測(transition detection) 語者預測(... | ||||
| 111 | 使用多情緒... 使用多情緒專家模型偵測新進語音情緒類別 (Adapt a New Emotion Class Detection by Speech using Mixture of Emotional Experts) | NTHU NDLTD | 碩 | 3.30 |
語音情緒辨識(speech emotion recognition)、多專家模型(mixture of experts)、小樣本學習(few shot learning) 語音情緒辨... | ||||
| 111 | 增強語音位... 增強語音位置隱私保護之權重對抗式去噪器 (A Weighted Adversarial Denoiser for Improving Speech Location Privacy Protection) | NTHU NDLTD | 碩 | 2.08 |
語音位置隱私(speech location privacy)、對抗式學習(adversarial learning)、去噪模型(denoise model)、權重對抗式去噪器(weighted adversarial denoiser(WAD))、語音情緒辨識(speech emotion recognition(SER))、語者驗證(speaker verification(SV)) 語音位置隱... | ||||
| 111 | 學習深度特... 學習深度特徵及標記空間以強化辨識任務 (Learning Deep Feature and Label Space to Enhance Discriminative Recognition Tasks) | NTHU NDLTD | 碩逕博 | 6.02 |
人本辨識任務(Human Centered Recognition)、任務分類方法(Task Categorization Scheme)、惡性血液疾病(Hematologic Malignancy)、流式細胞儀(Flow Cytometry)、情緒辨識(Emotion Recognition) 人本辨識任... | ||||
| 110 | 利用具情緒... 利用具情緒轉換感知之條件隨機場開發多模態對話中的情緒解碼系統 (Multimodal Emotion-Shift Aware CRF for Decoding Emotion Sequence in Conversation) | NTHU NDLTD | 碩 | 1.97 |
多模態情緒辨識(multimodal emotion recognition)、對話(conversation)、條件隨機場(conditional random field)、情緒轉換(emotion shift) 多模態情緒... | ||||
| 110 | 使用基於度... 使用基於度量學習之多任務網路建模電影中的角色關係與觀眾情緒 (Modeling In-Movie Character Relationship and Audience Emotion using Metric Learning based Multitask Network) | NTHU NDLTD | 碩(提早入學) | 2.43 |
電影資料庫(Movie Database)、角色關係(Character Relationship)、誘發情緒(Induced Emotion)、多任務學習(Multitask Learning)、三元組損失(Triplet Loss)、語文探索與字詞計算(Linguistic Inquiry and Word Count) 電影資料庫... | ||||
| 110 | 開發對談情... 開發對談情境中語談者交互過程計算框架 (Building Conversation-Oriented Interlocutors Interaction Process Modeling Framework) | NTHU NDLTD | 碩逕博 | 5.77 |
人類行為訊號處理(Behavioral Signal Processing)、對話(Conversation)、互動建模(Interaction Modeling)、表達行為(Expressive Behavior)、溝通功能(Communicative Function) 人類行為訊... | ||||
| 110 | 利用多聲學... 利用多聲學專家模型改進多模態電影場景分割 (Improving Multimodal Movie Scene Segmentation Using Mixture of Acoustic Experts) | NTHU NDLTD | 碩 | 2.72 |
電影(Movie)、場景分割(Scene Segmentation)、多專家模型(Mixture of Experts)、多模態注意力機制(Multimodal Attention)、聲訊(Audio) 電影(Mo... | ||||
| 110 | 使用人格特... 使用人格特質不變學習於基於生理訊號之情感辨識 (Physiological-based Affective Computing using Personality Invariant Learning) | NTHU NDLTD | 碩逕博 | 5.69 |
情緒識別(Emotion Recognition)、對抗式學習(Personality)、個體背景(Physiological Signals)、生理信號(Individual Context)、人格特質(Adversarial Learning) 情緒識別(... | ||||
| 110 | 運用對抗式... 運用對抗式學習網路結合多觀點輔助開發強健性聲音情緒識別 (Development of Robust Speech Emotion Recognition using Adversarial Learning Network with Multiple Perspective Auxiliary) | NTHU NDLTD | 碩逕博 | 6.64 |
跨資料庫情緒辨識(Cross-corpus emotion recognition)、對抗式學習(Adversarial learning)、轉移學習(Transfer learning)、領域適應(Domain adaptation) 跨資料庫情... | ||||
| 110 | 利用角色語... 利用角色語音行為預測小組任務成效進退步 (Team Progress Prediction using Role-based Vocal Behaviors in Small Group Interaction) | NTHU NDLTD | 碩 | 2.32 |
小組互動(small group)、語音行為(vocal behaviors)、小組進步(triplet loss) 小組互動(... | ||||
| 110 | 透過屬性對... 透過屬性對齊策略進行語音表徵學習 (Attribute-Aligned Speech Representation Learning Strategy) | NTHU NDLTD | 碩 | 2.26 |
語音表徵(speech representation)、隱私(privacy)、公平性(fairness)、屬性對齊(attribute alignment) 語音表徵(... | ||||
| 109 | 開發計算與... 開發計算與量化人類行為於個人文化維度與疼痛情緒之應用 (Toward Development of Computational Modeling and Quantifying Behaviors in Culture Dimension and Pain-Affect) | NTHU NDLTD | 碩逕博 | 6.25 |
行為訊號處理(Behavioral Signal Processing (BSP))、疼痛(Pain)、情緒(Emotion)、文化權力距離(Culture power distance)、功能性磁振造影(fMRI)、韻律學(prosody)、中心損失(center-loss embedding)、多頭注意力(Multi-Head Attention)、多任務學習(Multi-Task Learning) 行為訊號處... | ||||
| 109 | 基於條件式... 基於條件式相互學習之聯邦學習: 阿茲海默症在T1 加權磁振造影之分類辨識 (Federated Learning via Conditioned Mutual Learning for Alzheimer’s Disease Classification on T1-weighted MRI) | NTHU NDLTD | 碩 | 2.24 |
聯邦學習(Federated Learning)、知識蒸餾(Knowledge Distillation)、相互學習(Mutual Learning)、個性化學習(Personalized Learning)、阿茲海默症(Alzheimer's Disease)、磁振造影(Magnetic Resonance Image) 聯邦學習(... | ||||
| 109 | 利用表型特... 利用表型特徵嵌入注意力機制與資料擴增改善過動症 於功能性磁振造影中之預測 (Improve ADHD Classification in Rs-fMRI by Phenotypic-Attribute Attentional Embedding and Data Augmentation) | NTHU NDLTD | 碩 | 2.19 |
注意力不足過動症(Attention Deficit/Hyperactivity Disorder)、注意力機制(Attention mechanism)、資料增強(Data augmentation) 注意力不足... | ||||
| 109 | 圖像對話聲... 圖像對話聲學神經網絡： 在預測任務績效時對人格與聲學行為之間的小組內和小組間影響建模 (A Graph Interlocutor Acoustic Network: Modeling Intra-group and Inter-group Effects between Personality and Vocal Behavior in Predicting Task Performances) | NTHU NDLTD | 碩 | 2.05 |
小組互動(group interaction)、個性(personality)、注意力機制(attention mechanism)、圖卷積網絡(graph convolutional networks) 小組互動(... | ||||
| 109 | 從受眾反應... 從受眾反應探討多媒體對社群與個人影響: 電影票房預測和個人化情緒分析 (Learning Impact of Multimedia on Society and Individuals from Audience Reactions to Videos: Box Office Prediction and Personalized Affective Analysis) | NTHU NDLTD | 碩 | 2.03 |
多媒體(multimedia)、電影票房預測(box office prediction)、受眾反應分析(audience reaction analysis)、鑲嵌表徵學習(embedding learning) 多媒體(m... | ||||
| 109 | 利用自動語... 利用自動語音辨識特徵開發對話中的語音情感辨識系統 (ASR Dependent Speech Emotion Recognition in Spoken Dialog) | NTHU NDLTD | 碩 | 2.01 |
語音情感辨識(speech emotion recognition)、語音對話(spoken dialogs)、前後文(context)、注意力機制(attention mechanism)、情緒解碼(emotion decoding)、端到端語音辨識(end-to-end ASR)、領域自適應(domain adaptation) 語音情感辨... | ||||
| 108 | 利用多媒體... 利用多媒體資料建構的語音前端網路觀察情緒辨識重要資料 (Observe Critical Data in Emotion Recognition Using a Speech Front-End Network Learned from Media Data In-the-Wild) | NTHU NDLTD | 碩 | 2.68 |
語音情緒辨識(speech emotion recognition)、卷積神經網路(convolutional neural network)、語音前端網路(speech front-end network)、初始化微調(initialization fine-tuning) 語音情緒辨... | ||||
| 108 | 異質性資料... 異質性資料在人類行為其信息交互過程之探討:疼痛程度評估和多媒體暴力檢測 (Investigate Communication Process of Human Behavior from Heterogeneous Data: Pain Level Assessment and Multimedia Violence Prediction) | NTHU NDLTD | 碩 | 2.56 |
人類行為訊號處理(Behavioral signal processing)、疼痛程度辨識(pain level)、急診檢傷分類(triage)、傳媒暴力程度預測(violence prediction) 人類行為訊... | ||||
| 108 | 利用個人與... 利用個人與情境差異改善語音情緒辨識系統 (Improve speech emotion recognition system by considering individual and context difference) | NTHU NDLTD | 碩 | 2.07 |
語音情緒辨識(speech emotion recognition)、跨資料庫語音情緒分析(multi-speaker)、多語者模型(cross corpus SER) 語音情緒辨... | ||||
| 108 | 異質性表徵... 異質性表徵鑲嵌網絡對當代多媒體與主流文化之分析研究 (Heterogeneous Embedding Representation Network to Analyze Contemporary Multimedia and Prevailing Culture) | NTHU NDLTD | 碩 | 2.01 |
人類行為訊號處理(behavioral signal processing)、深度學習(multi-model)、社群媒體網絡分析(deep learning)、圖論捲積網絡(unsupervised learning)、多模態(graph convolutional network) 人類行為訊... | ||||
| 107 | 條件對比嵌... 條件對比嵌入網路: 透過元資訊學習輔助磁振造影特徵向量的學習 (A Condition-Contrastive Embedding Network: Using Meta Information to Guide MRI Representation Learning) | NTHU NDLTD | 碩 | 1.99 |
功能性磁振造影(fMRI)、特徵向量學習(representation learning)、人臉識別能力(face perception ability)、失智症(dementia) 功能性磁振... | ||||
| 107 | 雙重互補聲... 雙重互補聲學嵌入網絡: 從原始波形挖掘與特徵集相異的語音情感識別特徵 (A Dual Complementary Acoustic Embedding Network: Mining Discriminative Characteristics from Raw-waveform for Speech Emotion Recognition) | NTHU NDLTD | 碩 | 2.55 |
語音情緒辨識(speech emotion recognition)、原始波型(raw waveform)、端對端學習(end-to-end learning)、聲學空間擴大(acoustic space augmentation) 語音情緒辨... | ||||
| 107 | 利用長短期... 利用長短期記憶網絡之遺忘閘提取語音及文字流暢度特徵 用以改善自閉症孩童說故事自動辨識系統 (Learning Lexical and Speech Coherence Representation by Using LSTM Forget Gate) | NTHU NDLTD | 碩 | 2.13 |
人類行為訊號處理(behavioral signal processing)、語意流暢度(lexical coherence)、語音流暢度(speech fluency)、長短期記憶模型(long-short term memory neural network (LSTM))、自閉症(autism spectrum disorder (ASD))、說故事(story-telling) 人類行為訊... | ||||
| 107 | 基於大腦靜... 基於大腦靜息態迴旋積自編碼的fMRI特徵擷取器 (A fMRI Feature Extractor Based On Resting State Pretrained Convolutional Autoencoder) | NTHU NDLTD | 碩 | 2.06 |
fMRI 特徵擷取(fMRI feature extraction)、迴旋積自編碼(Convolutional Autoencoder)、大腦預設模式(default mode network) fMRI ... | ||||
| 106 | 透過表演逐... 透過表演逐字稿之互動特徵以改善中文戲劇表演資料庫情緒辨識系統 (Improving Chinese Performance Corpus Emotion Recognition System by Using Transcripts Interactive Features) | NTHU NDLTD | 碩(外籍生) | 2.89 |
行為訊號處理(behavior signal processing)、情緒辨識(emotion recognition)、自然語言處理(NLP)、支持向量機(SVM)、長短時記憶網路(long short-term memory) 行為訊號處... | ||||
| 106 | 利用多模態... 利用多模態模型混合CNN和LSTM影音特徵以自動化偵測急診病患疼痛程度 (Toward Automatic Pain-Level Detection for Emergency Patients using Fusion of CNN and LSTM Multimodal Audio-Video Features) | NTHU NDLTD | 碩 | 2.49 |
急診檢傷分類(Triage)、疼痛程度辨識(Pain recognition)、行為訊號處理(Behavior signal processing)、多模態融合(Multimodal fusion)、迴旋積類神經網路(Convolution neural network)、長短期記憶(Long short-term memory) 急診檢傷分... | ||||
| 106 | 以雙向長短... 以雙向長短期記憶網路架構混和多時間粒度文字模態改善婚 姻治療自動化行為評分系統 (Improving Automatic Behavior Rating System of Couple Therapy using Multi-granular Word Fusion Approach with bidirectional LSTM Architecture) | NTHU NDLTD | 碩 | 2.49 |
人類行為訊號處理(behavioral signal processing)、多時間粒度(multi-granularity)、深度學習(deep learning)、情感計算(affective computing)、自然語言處理(natural language processing) 人類行為訊... | ||||
| 106 | 利用LST... 利用LSTM演算法基於自閉症診斷觀察量表訪談建置辨識自閉症小孩之評估系統 (Using LSTM algorithm to establish an Evaluation System for Child with Autistic Disorder during Autism Diagnostic Observation Schedule Interview) | NTHU NDLTD | 碩 | 2.35 |
泛自閉症障礙(autism spectrum disorder)、自閉症診斷觀察量表(autism diagnostic observation schedule)、長短時記憶(long short-term memory)、人類行為訊號處理(behavioral signal processing)、多模態行為(multimodal behaviors) 泛自閉症障... | ||||
| 106 | 利用聯合因... 利用聯合因素分析研究大腦磁振神經影像之時間效應以改善情緒辨識系統 (Improving Categorical Emotion Recognition System by Joint Factor Analysis from Time Effect on fMRI) | NTHU NDLTD | 碩 | 2.19 |
語音情緒辨識(Speech emotion recognition)、功能性磁振造影(Functional magnetic resonance imaging)、時間(Time)、聯合因素分析(Joint Factor analysis)、激動程度(Activation)、情緒正負向(Valence) 語音情緒辨... | ||||
| 105 | 應用多任務... 應用多任務與多模態融合技術於候用校長演講自動評分系統之建構 (Toward Automatic Assessment of Pre-service Principals Oral Presentation using Multitask and Multimodal Fusion Technique) | NTHU NDLTD | 碩 | 無口試日期 |
人類行為訊號處理(behavioral signal processing)、教育研究(educational research)、口頭演講(oral presentation)、多模態訊號處理(multimodal signal processing)、多任務學習(multi-task learning) 人類行為訊... | ||||
| 105 | 基於多模態... 基於多模態主動式學習法進行樣本與標記之間的關係分析於候用校長評鑑之自動化評分系統建置 (An Multimodal Active Learning Approach in the Identification of Samples-to-Labels toward Developing Automatic Oral Presentation Assessment in the Pre-service Principals Certification Program) | NTHU NDLTD | 碩 | 無口試日期 |
主動式學習(active learning)、資料選取(data selection)、資料探勘(data mining)、機器學習(machine learning)、未標記資料(unlabeled data) 主動式學習... | ||||
| 105 | 針對實體化... 針對實體化交談介面開發基於行為衡量方法於自閉症小孩之評估系統 (Toward automatic assessment of child with autism using embodied conversational agents based on behavior-based measurement) | NTHU NDLTD | 碩 | 無口試日期 |
泛自閉症障礙(Autism spectrum disorder)、實體化交談介面(Embodied conversational agents)、自閉症診斷觀察量表(Autism diagnostic observation schedule)、人類行為訊號處理(Behavioral signal processing) 泛自閉症障... | ||||
| 105 | 透過結合f... 透過結合fMRI大腦血氧濃度相依訊號以改善語音情緒辨識系統 (Improving Categorical Emotion Recognition by Fusing Audio Features with Generated fMRI Brain Responses) | NTHU NDLTD | 碩 | 無口試日期 |
人類行為訊號處裡(behavioral signal processing(BSP))、情緒辨識(emotion recognition)、情緒正負向(valence)、功能性磁振造影(fMRI)、高斯混合回歸模型(Gaussian Mixture Regression) 人類行為訊... | ||||
| 105 | 一個多模態... 一個多模態連續情緒辨識系統與其應用於全域情感辨識之研究 (A Study on Automatic Multimodal Continuous Emotion Tracking and Its Application in Global Affect Recognition) | NTHU NDLTD | 碩 | 無口試日期 |
行為訊號處理(Behavior Signal Processing)、多模態情緒辨識(Multimodal Emotion Recognition)、薄片擷取(Sequence to Sequence) 行為訊號處... | ||||
| 105 | 結合fMR... 結合fMRI之迴旋積類神經網路多層次特徵 用以改善語音情緒辨識系統 (Improving Audio-based Categorical Emotion Recognition System by Fusing Convolutional Neural Network Hierarchical Features from fMRI Scanning) | NTHU NDLTD | 碩 | 無口試日期 |
fMRI(fMRI)、人類行為訊號處理(Behavior Signal Processing)、情緒辨識(Emotion Recognition)、迴旋積類神經網路(Convolutional Neural Network) fMRI(... | ||||
| 105 | 整合文本多... 整合文本多層次表達與嵌入演講屬性之表徵學習於強健候用校長演講自動化評分系統 (Enhancement of Automatic Assessment System for Pre-service Principals’ Oral Presentation using Speech Attribute-enriched Multi-level Feature) | NTHU NDLTD | 碩 | 1.97 |
行為訊號處理(behavior signal processing)、教育研究(educational research)、口頭演講(oral presentation)、自然語言處理(natural language processing)、演講屬性(self-defined attribute tag)、主題模型(topic model) 行為訊號處... | ||||
| 104 | 透過語音特... 透過語音特徵建構基於堆疊稀疏自編碼器演算法之婚姻治療中夫妻互動行為量表自動化評分系統 (Automating Behavior Coding for Distressed Couple Interactions Based on Stacked Sparse Autoencoder Framework using Speech-acoustic Features) | NTHU | 碩 | 無口試日期 |
深度學習(Deep Learning)、堆疊稀疏自編碼器(Stacked Autoencoders)、婚姻治療(Couple therapy)、人類行為分析(Human behavior analysis)、情緒分析(Emotion recognition) 深度學習(... | ||||
| 104 | 基於健保資... 基於健保資料預測中風之研究並以Hadoop作為一種快速擷取特徵工具 (Predicting Stroke based on Health Insurance Records by Using Hadoop as a Fast Feature Extraction Tool) | NTHU NDLTD | 碩 | 無口試日期 |
Hadoop(Hadoop)、GBDT(GBDT)、全民健康保險研究資料庫(National Health Insurance Research Database)、醫療衛生資料分析(analysis of healthcare data) Hadoo... | ||||
| 104 | 一個利用人... 一個利用人類Thin-Slice情緒感知特性所建構而成之全時情緒辨識模型新框架 (A Novel Global Affect Recognition Framework by Modeling Human’s Thin-Slice Emotion Perception) | NTHU NDLTD | 碩 | 無口試日期 |
行為訊號處理(Behavior Signal Processing(BSP))、thin-slice情感接收(thin-slice affect perception)、多模態情緒辨識(multimodal emotion recognition) 行為訊號處... |