電腦科學與資訊工程

利用電腦瀏覽圖形式資訊的時候，常受到螢幕空間大小的限制，沒有辦法在顯示圖形整體結構的同時也顯示細節部分。超廣角鏡頭是一種短焦距、大視角的相機鏡頭，鏡頭成像的時候，會有中間部分放大而周邊部分縮小的情形，藉由這個特性，我們發展出了一種新的圖形導覽介面，在瀏覽圖形式資訊的時候，有個圓形區域，該區域可隨著使用者的意願而自由移動，而區域內的圖形是以模擬超廣角鏡頭成像的方式呈現，且能夠與圓形區域外的圖形做銜接，如此，在瀏覽圖形式資訊的時候，除能夠顯示整體的結構外，也可以不開啟新視窗及無遮蔽的方式，即時地將想要觀察的部分做局部放大以展現細節。

Autonomous car is a very new concept, being a car without any driver. Several concurrent software process data using Artificial Intelligence to recognize and propose a path which the car should follow. The goal of the project is that a driverless car can reduce the distance between the cars, lowering the degree of road loadings, reducing the number of traffic jams, avoid human errors, and allowing people with disabilities(even blind people) to travel using an autonomous car. Theoretically a car without driver in the future should be much safer, because human reaction speed is higher than 200 ms, and the computing power of the newest computers allows traffic calculations even to 10 ms. The necessary power is provided by three multi-core laptops that process with Artificial Intelligence in order to recognize traffic signs, traffic lanes , traffic car fingerprints, processing the data from a 3D radar, using particle filters to localize car in a GPS map, the management of database with traffic signs, magnetic sensors, acceleration sensors, a distributed software, a supervisory system and the software which drives the stepper motor to turn the steering wheel (acceleration and braking). Currently the software is able to recognize the traffic signs, register them in a database using Google Maps. The fields record the sign and direction of travel from that area. Each car participating in the traffic and using this software will register new signs detected and the will modify the degree of confidence of recognition for other users. Another software component is able to recognize the demarcation lines between lanes, with three cameras to calculate exactly or using probabilities where it is on the road, where the roadsides are and to propose a new direction even in the absence of traffic signs for the next seconds. Another part of the software is trying to use Artificial Intelligence to detect other car fingerprints from webcam images. The calculation was performed on 3 computers, requiring distributed processing. I developed a management information system based on semaphores that allows data processing and supervision from 3 different computers. This project presents a hardware version of a LIDAR – a 3D radar and a software for creating a 3D environment in which the car navigates and using it the car will take decision to avoid obstacles. The LIRDAR contains a total of 16 avalanche photo-detector mounted on a stepper motor that spins at a frequency of 10 Hz. The information provided by my radar is about 576.000 pixels at resolution of 10 bits. The 3D radar helps the entire software system to increase the confidence of decision.

本研究提出一可用於機器人空間探勘與辨識的仿生機器學習系統。本系統模仿生物大腦的層級性結構，各層級間透過雙向連結進行搜尋辨識與提示，並記憶空間中的感官、場景和位置資訊，分別由以下部分構成： 1. 感官細胞：辨識特定感官輸入類別。 2. 場景基模細胞：組合具方向性的感官細胞數據。 3. 網格細胞：接收移動數據的內在座標系統。 4. 位置細胞：整合感官數據與空間數據、建立拓樸空間認知地圖。 經模擬實驗證明，本系統能在第一次探勘時建立空間認知地圖，並於再次造訪時成功匹配位置細胞進行定位。本系統有異地探勘、在複雜空間中進行路徑與任務規劃等廣泛應用。

本研究提出一個新的超通用、每邊w個端點的四邊形水分子形交換方塊(Water-Molecule-Shaped Switch Block; WMSB)架構，以應用在FPGA之多點連線（multipoint interconnection）和諸多交換網路的設計上。超通用交換方塊(HUSB)的領域中，Fan[2]提出當前唯一一個(4, w)-HUSB，但Fan’s (4, w)-HUSB所需的開關個數大約是6.3w個開關，在接下來的篇幅之中，我們將證明(4, w)-WMSB是只需6w個開關的HUSB；此外，我們還證明沒有(4, w)-HUSB可以使用小於6w個開關。本研究中還使用VPR（一種CAD）及其內建的大量標準線路以證明(4, w)-WMSB不僅是理論上最佳的亦是實用性佳的交換方塊。鑑此，(4, w)-WMSB開關效率高(switch-efficiency)的設計十分適用於其他的交換網路設計，如公共電話網路(Public Switched Telephone Network)。

Monte Carlo methods are non-parametric algorithms that use random numbers and theorems of probability theory to approximate values that are not random. The purpose of my research was to approximate the surface of different geographical areas that can be easily approximated to polygons (e.g. lakes, glaciers, deserts) with Monte Carlo simulations starting from either Cartesian coordinates or pictures. Computer science would not exist without math, and this research project showed me the importance of a deep understanding of probability theory in the world of simulations and, more generally, the importance of developing new theorems and algorithms. The results of my research could be developed in different ways: it would be interesting to produce software that allows one to approximate areas from pictures taken from a smartphone; as well, the theorem I found has to be proven, and also Monte Carlo methods as a means of random number generation can always be improved. There are still many possibilities.

以製作適合高中生的英文篇章難易度自動分級為初衷，本研究採高中英文課文為語料，針對「如何分級」，意即從文章萃取哪些特徵、利用何工具或語料協助萃取特徵、以何工具分級等因素，進行研究與實驗，並建立一套新方法。首先進行前處理，再嘗試以單字、句型的數量或比例、句長、音節長、整合以上分析等各式特徵，支持向量機(Support Vector Machines)、隨機森林分類器(Random Forest Classifier)、決策樹分類器(Decision Tree Classifier)、卷積神經網路句分類器(Convolutional Neural Networks for Sentence Classification)等工具，進行將篇章分為高中一、二、三年級等三個難易度等級的測試，建立自動分級模型。最後製作成可供大眾使用的自動分級網頁。各項測試之中，最佳分類效能為整合各項特徵時得到的分類正確率65.04%，經模擬得知，此效能較過去研究，已有所提升。

我們每天的生活都離不開手機、電腦、電視等產品，因此藍光對眼睛造成的影響是所有人都會遇到的問題。本計劃希望能利用人眼對於不同顏色的敏感度，在人眼感知範圍內減少螢幕的藍光強度，以降低電子產品對眼睛帶來的負擔及傷害，同時維持螢幕畫面的正常顯色。市面上現有的方法除了成本較高，也都會使螢幕畫面變得昏暗; 若為了維持顯色的自然，則成效便會受到限制。本專題根據人眼對於色度的最小可覺差，參考基於麥克亞當橢圓的顏色差異計算方式，分別算出不同色彩空間上看起來相同的顏色關係表，並將螢幕上的顏色換為色差無法察覺且藍色強度較低的顏色，最後以手持式光譜儀量測螢幕輸 出的藍光強度驗證成效。與現行市面上的抗藍光方法相比，除了成本大幅降低以外，也能維持顯示器的正常顯色。未來希望能推廣至所有 LED 螢幕電子產品，並為色盲色弱患者設計不同色度感知能力的最佳調整方式。

Millions of people worldwide suffer from aphasia, a disorder that severely inhibits language comprehension. Medical professionals suggest that individuals with aphasia have a noticeably greater understanding of pictures than of the written or spoken word. Accordingly, we design a text-to-image converter that augments lingual communication, overcoming the highly constrained input strings and predefined output templates of previous work. This project offers four primary contributions. First, we develop an image processing algorithm that finds a simple graphical representation for each noun in the input text by analyzing Hu mo-ments of contours in images from The Noun Project and Bing Images. Next, we construct a da-taset of 700 human-centric action verbs annotated with corresponding body positions. We train support vector machines to match verbs outside the dataset with appropriate body positions. Our system illustrates body positions and emotions with a generic human representation created using iOS’s Core Animation framework. Third, we design an algorithm that maps abstract nouns to concrete ones that can be illustrated easily. To accomplish this, we use spectral clustering to iden-tify 175 abstract noun classes and annotate these classes with representative concrete nouns. Fi-nally, our system parses two datasets of pre-segmented and pre-captioned real-world images (Im-ageClef and Microsoft COCO) to identify graphical patterns that accurately represent semantic relationships between the words in a sentence. Our tests on human subjects establish the system’s effectiveness in communicating text using im-ages. Beyond people with aphasia, our system can assist individuals with Alzheimer’s or Parkin-son’s, travelers located in foreign countries, and children learning how to read.

量子計算的效率相較傳統計算有指數級成長。然而此領域中多數研究皆專注於量子計算的性質本身，鮮少討論如何將傳統環境中的既有資訊轉換至量子環境下。一旦量子電腦實現，受量子效應限制，傳統資料多半不能相容於量子環境中。因此，本研究的目的是發想出一種系統性的演算法以確實跨越量子資訊與傳統資訊之間資料結構的藩籬。我們選擇的計算機模型是確定有限狀態自動機（Deterministic Finite Automaton，簡稱DFA）。 本研究由自動機的轉移矩陣（Transition Matrix）及量子環境要求的可逆性（Reversibility）出發，自傳統DFA一步步轉換至量子有限狀態自動機（Quantum Finite Automaton，簡稱QFA）並進行優化。最終，我們定義出一種新的QFA模型（QDFA）能在量子環境下運行，具有增大的字母表（Alphabet Set）但功能完全等價於DFA（能辨認正則語言）。本研究獨創的演算法的時間複雜度為O(C×N2)。

大數據時代促成運動賽事分析更加蓬勃發展，一般來說數據分析需要三個要素：資料、分析演算法和應用領域知識；隨著開放資料分享的普遍化，人們可以更輕易的取得資料並運用各式分析演算方法來針對有興趣的應用問題做出更精準的預測。 本研究以kaggle平台上所提供之歐洲職業足球比賽之公開資料集為基礎，使用目前最具分析潛力的深度學習技術─結合卷積神經網路（CNN）和全連接型神經網路的設計出一個五層的學習架構，建置出足球比賽結果的預測模型。此模型可直接預測主隊勝、負以及平手等三種結果，實驗結果亦展示出本研究所建置的SoccerNet預測模型優於過往的研究，有著更佳的預測能力；同時也驗證了使用公開資料集與CNN技術在球賽分析的可能性。而本研究所提的SoccerNet模型不僅可以運用於賽前的結果預測，亦可運用於球隊經營管理等決策，頗具有商業價值。