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(年) |
|---|---|---|---|---|
| 關鍵字 | 口試日期 | |||
| 113 | 結合人體溫... 結合人體溫度一致性增強 CycleGAN 在夜間熱影像上的生成 (Enhancing Nighttime Thermal Image Generation with CycleGAN via Human Body Temperature Consistency) | NTHU NDLTD | 碩 | 1.99 |
. .... | 2025-07-25 | |||
| 113 | 影像框級標... 影像框級標註錯誤修正之初期探討 (A Preliminary Study of Correcting Noisy Image Box-level Labeling) | NTHU NDLTD | 碩 | 1.99 |
. .... | 2025-07-25 | |||
| 113 | 基於對數報... 基於對數報酬率與多尺度波動特徵於碳期貨長期價格預測 (Long-term Carbon Futures Price Forecasting via Log-Return and Multiscale Volatility Features) | NTHU NDLTD | 碩 | 1.98 |
. .... | 2025-07-23 | |||
| 113 | 結合多尺度... 結合多尺度波動特徵之碳價長期預測模型 (Long-term Carbon Price Forecasting with Multiscale Volatility Features) | NTHU NDLTD | 碩(提早入學) | 1.48 |
. .... | 2025-07-23 | |||
| 113 | 基於單一感... 基於單一感測器之拳擊力道預測：結合物理啟發混合專家模型 (Punch Force Estimation from a Single Sensor Using a Physics- Guided Mixture-of-Experts Framework) | NTHU NDLTD | 碩 | 1.96 |
. .... | 2025-07-15 | |||
| 113 | 基於視覺區... 基於視覺區塊重要性建構無標註影像分割核心集 (Annotation-Free Coreset Selection for Image Segmentation by Visual Patch Importance) | NTHU NDLTD | 碩 | 1.96 |
. .... | 2025-07-15 | |||
| 112 | 利用幾何變... 利用幾何變換進行自適應模糊災害影像分析 (BRIDGE: Blur Adaptive Disaster Images Analysis by Geometric Transformation) | NTHU NDLTD | 碩(外籍生) | 2.00 |
災害影像分析(Disaster Images Analysis)、模糊自適應空間變換(Blur Adaptive Spatial Transformation)、AI模型銜接(AI Models Bridging)、緊急應變系統(Emergency Response Systems)、深度學習(Deep Learning) 災害影像分... | 2024-07-31 | |||
| 112 | Drop2... Drop2Sparse: 透過稀疏模型改善資料集蒸餾技術 (Drop2Sparse: Improving Dataset Distillation via Sparse Model) | NTHU NDLTD | 碩 | 2.00 |
資料集壓縮(Dataset Compression)、資料集蒸餾(Dataset Distillation)、梯度配對(Gradient Matching)、模型稀疏(Model Sparsification) 資料集壓縮... | 2024-07-31 | |||
| 112 | 基於CAM... 基於CAM引導GrabCut進行弱監督皮膚病變影像分割 (CAM-Cut: Weakly Supervised Skin Lesion Segmentation by CAM Guided GrabCut) | NTHU NDLTD | 碩 | 2.00 |
類激活圖(Class Activation Map)、深度學習(Deep Learning)、皮膚病變(Skin Lesion)、弱監督語意分割(Weakly Supervised Semantic Segmentation) 類激活圖(... | 2024-07-31 | |||
| 112 | 以CAM引... 以CAM引導的高斯計分板進行方框層級之資料標記的品質評估 (Quality Assessment of Box-level Data Labeling by CAM guided Gaussian Scoreboard) | NTHU NDLTD | 碩 | 1.99 |
類激活圖(Class Activation Map)、資料標記(Data Labeling)、深度學習(Deep Learning)、標籤品質衡量(Labeling Quality Assessment) 類激活圖(... | 2024-07-25 | |||
| 112 | 以密度為基... 以密度為基礎的區域選取與特徵融合進行專利影像檢索 (Density-Refine: Patent Image Retrieval by Density based Region Extraction and Feature Fusion) | NTHU NDLTD | 碩 | 1.99 |
機器學習(machine learning)、機器視覺(computer vision)、檢索(retrieval) 機器學習(... | 2024-07-25 | |||
| 112 | 雙極擴增:... 雙極擴增: 察覺色偏情境的輕量級異常檢測模型訓練 (Bipolar Augmentation: Lightweight AI Model Training for Anomaly Detection under Color Cast Scenarios) | NTHU NDLTD | 碩 | 2.14 |
深度學習(deep learning)、色偏(color cast)、異常偵測(anomaly detection)、資料擴增(data augmentation) 深度學習(... | 2023-09-21 | |||
| 111 | 模組化學習... 模組化學習：於穩健性交通號誌辨識的高效漸進式開發 (Modular Learning: Efficient Incremental Development of Robust Traffic Sign Recognition) | NTHU NDLTD | 碩 | 1.97 |
交通號誌辨識(Traffic Sign Recognition)、AI訓練邊際成本(Marginal Cost of AI Training)、知識蒸餾(Knowledge Distillation)、圖像毀損(Image Corruption) 交通號誌辨... | 2023-07-19 | |||
| 111 | 從戴口罩的... 從戴口罩的影像資料訓練深偽檢測模型 (Training Deepfake Detection Model from Photos with Face Mask) | NTHU NDLTD | 碩 | 1.97 |
深偽檢測(Deepfake Detection)、口罩(Face Mask)、訓練策略(Training Strategy)、局部合成(Partial Blending) 深偽檢測(... | 2023-07-19 | |||
| 111 | 利用複合式... 利用複合式縮放強化異常聲音檢測模型之性能 (Extended STgram: Improved Anomalous Sound Detection by Compound Scaling) | NTHU NDLTD | 碩 | 1.95 |
異常聲音檢測(Anomalous Sound Detection)、自監督學習(Self-Supervised Learning) 異常聲音檢... | 2023-07-12 | |||