https://ir.lib.ncu.edu.tw/handle/987654321/40 Search the Channel s https://ir.lib.ncu.edu.tw/simple-search https://ir.lib.ncu.edu.tw/handle/987654321/111464 title: 融入座標注意力與FiLM機制之條件式擴散模型於射出成型光彈圖預測;Conditional Diffusion Model Incorporating CoordinateAttention and FiLM Mechanisms for Injection MoldingPhotoelastic Fringe Image Prediction abstract: 「合作論證學習」是一種以小組互動與論點建構為核心的學習方式，主張學生透過討論、論點交流、證據提出與反駁修正，一步一步共同建構知識與深化理解的過程。這種學習方式有助於提升學生的批判思考能力、問題解決能力與論述表達能力。而其中，英文議論文（argumentative essay）被視為最能體現學生批判思考與論證能力的文類之一。英文議論文不僅要求學生具備基本語言能力，更需要學生能清楚界定立場、組織論點、提出證據支持立場，並回應反方觀點，以展現邏輯推理與批判思維。同時，隨著近年來資訊科技迅速發展，人工智慧已逐步被廣泛應用於各個領域，在教育領域中，以聊天機器人為基礎的學習工具開始被視為能夠協助學生學習、提供即時回饋的學習夥伴。 因此本研究導入人工智慧技術，設計能自動摘要分類小組論點、證據以及引導學生逐步將組內論證成果轉化為完整英語議論文寫作大綱之「AI聊天機器人」，協助學生在寫作前進行論點組織、論證深化與結構化思考，進一步協助學生從合作論證轉向有效的個人英文議論文寫作。 本研究採系統初步評估設計，針對大一學生進行實測。結果顯示，學生對平台之知覺有用性、知覺易用性與使用意願皆持正向看法，並肯定平台在協助英文議論文寫作規劃、論證思考引導與架構釐清方面之成效。研究結果初步顯示，結合 Knowledge Forum 合作論證活動與 AI 寫作引導之平台，具有協助學生整理小組論證內容、形成個人化寫作素材，並支援英文議論文寫作前規劃與架構建構之潛力。;Collaborative Argumentation Learning is a learning approach centered on group interaction and argument construction， in which students collaboratively build knowledge through discussion， evidence sharing， and rebuttal. This approach helps enhance students critical thinking， problem-solving， and argumentative expression abilities. Argumentative essay writing， which emphasizes stance development， evidence support， and logical reasoning， is regarded as an important way to cultivate students critical thinking and argumentation skills. Meanwhile， with the rapid development of artificial intelligence technologies， chatbot-based learning tools have gradually been adopted in education as learning partners that provide immediate feedback and support. Therefore， this study integrates artificial intelligence technology to develop a chatbot that can automatically summarize and categorize group arguments and evidence， while guiding students to transform collaborative discussion outcomes into structured English argumentative essay outlines， thereby supporting students argument organization and writing development. This study adopted a preliminary system evaluation design and conducted a pilot implementation with first-year university students. The results indicated that students held positive perceptions regarding the platform’s perceived usefulness， perceived ease of use， and intention to use. Students also recognized the platform’s effectiveness in supporting argumentative essay planning， guiding argumentative thinking， and organizing writing structures in English argumentative writing. The findings suggest that the platform has the potential to support students in transforming group discussion content into personalized writing materials and in planning the structure of English argumentative essays. <br> Tue, 20 Oct 2026 10:17:52 GMT https://ir.lib.ncu.edu.tw/handle/987654321/111462 title: 差動式雙質量塊結構之二軸 MEMS 磁力計設計;Design of a Differential Dual-Mass Structure for a Two-Axis MEMS Magnetometer abstract: 本研究提出一種透過雙質量塊差動模式之二軸 MEMS 磁力計，透過雙 質量塊反向作動之設計，有效提升感測靈敏度與輸出訊號品質。元件利用勞 倫茲力驅動，將外加磁場轉換為微結構位移，並進一步透過差分電容變化轉 換為電壓訊號。此磁力計使用UMC 0.18 m CMOS製程，結構尺寸分別為 為1020 × 994 ? <br> Tue, 20 Oct 2026 10:10:20 GMT https://ir.lib.ncu.edu.tw/handle/987654321/111454 title: Reliable Artificial Intelligence Systems for Alzheimer′s Disease: Advanced Methods for Detection and Classification;Reliable Artificial Intelligence Systems for Alzheimer′s Disease: Advanced Methods for Detection and Classification abstract: 阿茲海默症的磁振造影(MRI)診斷雖已達到高分類準確率，但臨床可用性還要求模型能在真實病患的解剖變異下保持泛化，並能回報預測是否可信。本論文提出四項貢獻。第一，建立由特徵位置不變性(FPI)驗證、生物標記位置擴增(BLA)與高信賴度隊列(HCC)篩選構成的可靠性框架：FPI顯示海馬迴偵測器在位置改變後即失效，BLA以位置重排資料訓練解決此問題，HCC則以複合信賴分數移除低信賴度樣本，使MobileNet在AD對NC分類的準確率由96.75%提升至98.95%。第二，提出配對機率差異可靠性量化(PPDRQ)及其自訂損失函數，產生樣本專屬的信賴度分數；採用三元組損失時準確率達85%，期望校準誤差為0.05，與Deep Ensemble統計上等效(McNemar′s p = 0.94)，運算成本僅約其五分之一，並在DermNet上通過跨領域驗證。第三，設計結合一維LSTM海馬迴強度編碼器與二維CNN的雙串流框架，以加性注意力融合，驗證準確率達94.65%，優於二維基準(93%)。第四，提出混合CNN-GCN架構，在四種骨幹網路上皆優於僅使用CNN的基準(提升0.7至2.2個百分點)，以ResNeSt-50達88.5%準確率，並經GNNExplainer定位海馬迴與內嗅皮質。整體而言，本研究將準確率、可靠性與可解釋性視為彼此耦合、各自驗證的需求，為可稽核的AI輔助阿茲海默症診斷提供可行途徑。;Alzheimer′s disease diagnosis from MRI has reached high classification accuracy in research settings， but accuracy alone does not establish clinical readiness. Two further properties matter: the model must generalize across the anatomical variation seen in real patients， and it must report when a prediction should be trusted versus flagged for review. This thesis develops four contributions that target these properties directly. The first is a reliability framework built from Feature Position Invariance (FPI) validation， Biomarker Location Augmentation (BLA)， and a High Confidence Cohort (HCC) filtering pipeline. FPI testing shows that hippocampus detectors trained on standard scans fail once the hippocampus is repositioned， exposing a dependence on spatial location rather than tissue features. BLA counters this by training on repositioned data， and HCC removes low-confidence training samples using a composite score combining softmax probability， augmented-inference consistency， and expert labeling. HCC-filtered training raises MobileNet accuracy on AD-versus-NC classification from 96.75% to 98.95%. The second contribution is Pairwise Probability Differential Reliability Quantification (PPDRQ)， an uncertainty metric paired with a custom loss function that produces a sample-specific， class-resolved confidence score. Benchmarked against MC-Dropout and Deep Ensemble， PPDRQ with triplet loss reaches 85% accuracy with an expected calibration error of 0.05， statistically equivalent to Deep Ensemble (McNemar′s p = 0.94) at roughly one-fifth the computational cost. Cross-domain validation on DermNet confirms the metric transfers outside brain MRI， with high-confidence predictions reaching 91–92% accuracy. The third contribution is a dual-stream diagnostic framework combining a 1D LSTM encoder for hippocampal intensity signals with a 2D CNN for whole-brain anatomical context， fused through additive attention. This framework reaches 94.65% validation accuracy， improving on a 2D-only baseline (93%) with each added stream. The fourth contribution is a hybrid CNN-GCN architecture that converts CNN feature maps into a graph of spatial patches connected by both adjacency and feature similarity， with complexity-aware node encoding that learns diagnostic importance per patch. This graph module improves over CNN-only baselines across four backbones (+0.7 to +2.2 percentage points)， reaches 88.5% accuracy with a ResNeSt-50 backbone， and produces GNNExplainer visualizations that localize to hippocampal and entorhinal regions without hand-coded anatomical priors. Together， these contributions treat accuracy， reliability， and interpretability as coupled， separately validated requirements rather than a single aggregate metric， pointing toward AI-assisted AD diagnosis that clinicians can audit rather than take on faith. <br> Tue, 20 Oct 2026 09:36:10 GMT https://ir.lib.ncu.edu.tw/handle/987654321/111453 title: 液體輔助飛秒雷射藍寶石晶圓微鑽孔研究;Research on Liquid-Assisted Micro-Drilling of Sapphire Wafer Using Femtosecond Laser abstract: 藍寶石（Sapphire）為單晶的三氧化二鋁（Al2O3），具有高硬度、高化學穩定性、良好高溫穩定性及優異的光學特性等特點，使其廣泛應用於發光二極體（LED）與半導體基板、光電元件及精密封裝等領域。然而，由於藍寶石高硬度伴隨之高脆性，使得傳統接觸式加工於微鑽孔中面臨容易產生裂紋及工具磨耗等問題，為符合高效率與高精度之製程需求，非接觸式之飛秒雷射精密微鑽孔技術成為極具潛力之替代方案。 本研究以波長1030 nm、脈衝寬度450 fs之飛秒振鏡雷射系統，於試片下表面導入純水進行藍寶石晶圓背向鑽孔研究，目標在於透過將雷射焦點自試片上表面調整至下表面並搭配負離焦策略，將加工模式從由上往下燒蝕轉變為由下往上燒蝕，避免等離子屏蔽效應，從而抑制錐角孔形之生成，獲得孔壁較為垂直之孔洞；並透過液體輔助克服於空氣環境下鑽孔時殘屑排除及再鑄層堆積等問題，提升鑽孔深度與效率。在機制探討方面，本研究利用高速微距攝影技術，於由上往下鑽孔時捕捉到燒蝕羽流引發之等離子屏蔽效應及雷射光束於孔壁之折射與反射現象；並於液體輔助加工中，觀測記錄氣泡之生成潰滅動態及通孔瞬間所噴發之水柱，藉此瞭解加工過程中雷射光、材料與輔助液體三者之交互作用。對於厚度150 μm之藍寶石晶圓，可於0.566秒完成直徑50 μm，粗糙度0.668 μm之通孔。最後使用硫酸與磷酸混合溶液進行濕式蝕刻後處理，進一步去除加工過程因雷射熱影響於孔壁造成的改質區，透過TEM檢測結果證實，蝕刻後孔壁餘剩區域皆呈現排列整齊之藍寶石單晶結構。藉此，本研究成功實現了直徑50 μm，粗糙度0.449 μm，錐角小於1°，深寬比為3之圓形通孔。;Sapphire is the single-crystal form of aluminum oxide (Al₂O₃)， possesses high hardness， excellent chemical stability， good high-temperature stability， and superior optical properties， which have led to its widespread application in light-emitting diodes (LEDs)， semiconductor substrates， optoelectronic components， and precision packaging. However， the high brittleness accompanying the high hardness of sapphire causes conventional contact-based machining to suffer from crack formation and tool wear during micro-drilling. To meet the demands for high-efficiency and high-precision processing， non-contact femtosecond laser precision micro-drilling has emerged as a highly promising alternative. In this study， a femtosecond galvanometer laser system with a wavelength of 1030 nm and a pulse duration of 450 fs was employed to investigate the backside drilling of sapphire wafers， with pure water introduced onto the bottom surface of the specimen. By shifting the laser focal point from the top surface to the bottom surface of the specimen and applying a negative-defocus strategy， the machining mode was transformed from top-down ablation to bottom-up ablation， thereby avoiding the plasma shielding effect， suppressing the formation of tapered hole profiles， and obtaining holes with more vertical sidewalls. Furthermore， the liquid-assisted approach was adopted to overcome the problems of debris removal and recast layer accumulation encountered during drilling in air， thereby enhancing the drilling depth and efficiency. Regarding the mechanism investigation， high-speed micro-photography was utilized to capture， during top-down drilling， the plasma shielding effect induced by the ablation plume as well as the refraction and reflection of the laser beam at the hole wall; in the liquid-assisted process， the formation and collapse dynamics of bubbles and the water jet ejected at the instant of through-hole formation were observed and recorded， thereby elucidating the interactions among the laser beam， the material， and the assisting liquid during machining. For a sapphire wafer with a thickness of 150 μm， a through-hole with a diameter of 50 μm and a roughness of 0.668 μm could be completed within 0.566 seconds. Finally， a wet-etching post-treatment using a mixed solution of sulfuric acid and phosphoric acid was applied to further remove the modified region formed on the hole wall due to laser-induced thermal effects; TEM examination confirmed that the remaining region of the hole wall after etching exhibited a well-ordered single-crystal sapphire structure. Through this approach， this study successfully achieved a circular through-hole with a diameter of 51 μm， a roughness of 0.449 μm， a taper angle of less than 1°， and an aspect ratio of 2.94. <br> Tue, 20 Oct 2026 09:35:53 GMT