Chung Cheng University Institutional Repository
Not a member yet
889 research outputs found
Sort by
基於效能計數器的非揮發混合式記憶體快取管理優化演算法;Optimization Cache Management Algorithm of Non-Volatile Hybrid Memory Based on Performance Counter
No full text[[abstract]]相變化記憶體 (Phase-Change Memory, PCM) 因為具有非揮發性、相對於 DRAM 高 2 ~ 4 倍的儲存密度、極低的靜態功率消耗、較 Flash Memory 為高的寫入壽命與接近動態隨機存取記憶體 (Dynamic Random Access Memory, DRAM) 的存取速度等特性而備受矚目，有望在未來取代 DRAM 等揮發性記憶體成為主記憶體材料。但因為在現階段製程上 PCM 還存在著為了改變相態而需要較高的電壓產生足夠溫度、較高的寫入延遲，與寫入磨耗的考量等因素，故短時間內要完全取代 DRAM 估計是較難達成的。現今研究趨向於結合兩者的優點，因 DRAM 無限制寫入次數，將 DRAM 作為一個 PCM 的前級快取記憶體，能提高系統反應與透過降低 PCM 寫入次數來增加其壽命。本論文提出的快取管理演算法考量到揮發性記憶體、非揮發性記憶體的壽命、存取速度等特性，使用 CPU 內部效能計數器 (Performance Counter) 所提供的頁面存取頻率資訊對記憶體頁面做分類，進而決定頁面該被放置於何種記憶體材料中能滿足對效能及壽命的最佳化考量。
Phase-Change Memory (PCM), due to its many distinctive features: non-volatile, 2~4 times higher storage density than Dynamic Random Access Memory (DRAM), very low static power consumption, and orders of magnitude better than NAND Flash Memory in read latency, write latency, and write endurance, is expected to replace DRAM and other volatile memory to become the main memory material in the near feature. However, at this stage, the process of PCM manufacturing still has many unfavorable factors on matters such as to change its phase, in need of a higher voltage to produce enough temperature, high write latency and write wear consideration; the complete replacement of DRAM is therefore difficult in a short time. Research today tends to combine the advantage of both, with respect that DRAM has a close to infinite endurance time and its lower access latency; taking DRAM as a pre-cache memory of PCM can improve the system response and increase PCM life span by reducing the number of PCM write.The cache management algorithm proposed in this paper takes information about access frequency of memory page from CPU internal performance counter, consider endurance, access latency and other characteristics to classify memory pages, that then determines which memory material is to be placed to meet the performance and life span of the optimal considerations
LTE-A雙重連線異質網路中具基地台負載感知的基地台選擇機制;Cell Load-aware Cell Selection Mechanism in LTE-A Heterogeneous Networks with Dual Connectivity
No full text[[abstract]]隨著數據傳輸量的增長，小型基地台在長期演進技術成為重要的角色，歸功於小型基地台低功率和佈建成本低的特性，大型基地台的收訊死角或邊緣地區都能透過佈署小型基地台獲得改善，由小型基地台和大型基地台構成的異質網路，因兩者運行於同一個頻段，因此覆蓋範圍重疊時會出現干擾的情況。第三代合作夥伴計劃發表Release 12提出不同頻及雙重連線的概念，雙重連線指的是使用者裝置同時與大型基地台和小型基地台連線，處於基地台邊緣的裝置能藉此增加傳輸效能，大型基地台覆蓋範圍廣，若能與大型基地台保持連線，也可增進移動的穩固性，而且小型基地台使用比大型基地台更高的頻帶，就不會有干擾的問題。為了舒緩熱點區域網路壅塞的問題，可增加小型基地台的佈署密度，由數個小型基地台集合而成的稱為小型基地台叢集，當佈建密度上升時，必須有合適的方法來避免乒乓效應。本論文提出費用函數，為具有感知基地台負載狀況的服務基地台選擇機制，常見的方法多以使用者裝置接收到的傳輸功率強弱，作為選擇的依據。但訊號最好的基地台不一定能提供最好的服務品質，尤其當網路壅塞，基地台剩餘資源不足以提供使用者需求，因此我們能透過費用函數計算使用者由此基地台進行服務的費用，判斷出最佳的服務基地台。最後，我們將不同頻、雙重連線與提出的選擇機制作整合，並設計模擬情境來評估其表現，模擬結果顯示費用函數能有效促進基地台的負載平衡，並有良好的效能表現。
As the demand of data traffic grows, small cell has become a key solution to boost network efficiency. Characterized by its lower power consumption and less deployment cost, small cell can be massively deployed to the cell edge and coverage holes of macro stations, increasing the capacity of network. However, high interference can be expected if the macro cells and the small cells share the same spectrum. To cope with the situation, 3GPP Release 12 proposed the concept of dual connectivity. By dual connectivity, user equipment (UE) can connect to both a macro cell and a small cell simultaneously to increase network efficiency. Edge UEs can increase the robustness of mobility by connecting to the macro cell. In addition, to reduce interference between macro cell and small cells, the frequency (spectrum) of macro cell may be different from that of small cells.To solve the network congestion issue in the hotspot area, a typical solution is to increase the density of small cell deployment. Nearby small cells may formed a cluster. The conventional cell selection model is determined by signal power like reference signal received power (RSRP) or signal-to-interference-plus-noise ratio (SINR). With the higher density of the small cell deployment, a UE may easily handoff from one small cell to another one, causing the ping-pong effect. Furthermore, despite having better signal, the base station might unable to provide enough resource to its associated UEs when congestion happens. In this paper, we proposed a cell selection mechanism based on a cost function which is used to evaluate the traffic load of cells. Simulation results show that our proposed algorithm can improve load balance and conduct good performance in comparison with conventional studies
使用影像與文字對齊技術摘要部落格文章;A Study on Summarizing Blog Articles by Using Image-Text Alignment
No full text[[abstract]]現在部落格文章資訊越來越豐富，但是對於想快速地進行瀏覽或使用行動裝置瀏覽的使用者是相當不方便的，使用者常須不斷捲動網頁進行瀏覽。在本論文中，我們要進行部落格文章摘要來幫助這些使用者。我們觀察到以前的文章摘要技術幾乎都沒有參考影像資訊，當一篇文章有影像時，影像豐富文章內容，也提供重要的資訊。所以，我們將利用影像搭配文字資訊來做文章摘要，大大增加摘要的可看性，且大幅減少使用者的捲動行為。這裡我們利用近幾年發展相當快速的深度學習技術來實作影像與文字對齊技術的文章摘要系統。 首先，偵測出影像中許多可能是物件的區塊，每個物件區塊利用卷積神經網路萃取視覺特徵。文字部分先利用word2vec取文字特徵，並透過適當的轉換。之後再將視覺及文字兩種不同的特徵投影到相同的特徵空間，當文字和影像在相同特徵空間時，我們就可以直接計算兩者間的距離，透過這個距離我們可以知道影像與文章中的哪個句子語意相近，最後選出最適當的影像及文字配對形成文章摘要結果。 在實際的部落格實驗中，我們的方法勝過直接以image captioning結果做文章摘要，還有基於News2Image概念的文章摘要。我們也試著考慮文章與影像的空間特性加強文章摘要結果，並探討未來可再加強那些部分。
Rich content appears in blog articles currently. However, lengthy articles impede efficient access and often prevent users from reading articles from mobile devices. We thus would like to summarize blog articles to facilitate mobile browsing. Although images are often embedded in blog articles, previously document summarization works seldom consider image information. In this thesis, we would like to jointly consider image and text information, and summarize blog articles into image-sentence pairs so that readability can be enhanced. We will take advantage of deep learning technologies to achieve image-text alignment. Object regions in an image are first detected. From each object region we extract CNN features. On the other hand, text in sentences is transformed into vectors by the word2vec module. We then map image features and text features into a common space, where we can compute distance between sentences and images. Finally, semantics-related image-sentence pairs are determined to form blog article summaries. In the evaluation, we show that the result of our system is better than directly calculating similarity between sentences and image captions automatically generated by an image captioning engine. Our performance is also superior to a similar system, News2Image. We also try to consider spatial information of images to improve performance, and discuss future works
C?ab: 微生物自動序列分析之網路平台;C?ab: A web server for automated microbial sequence analysis
No full text[[abstract]]近幾年來，院內臨床醫生為患者建立抗生素處方面臨著巨大的挑戰。多重抗藥性細菌的產生，導致臨床醫生迫切需要在極短時間內檢測細菌的耐藥性，全基因體定序能分析細菌基因體進而得知其耐藥性。然而分析全基因體定序資料需要整合許多軟體工具與資料庫，且使用者介面普遍皆不甚友善。為此，我們開發了C?ab網絡服務，一種用於細菌基因組分析的線上微生物序列分析平台。C?ab統整目前院內臨床醫生針對微生物菌株的鑑定標準檢測流程，其中包含下列四項主要功能：三代定序技術之基因體組裝、物種鑑定、毒性基因與抗藥基因的檢測。此外，用戶可以透過網頁瀏覽器手動設置參數或選擇預設參數，將分析的前置作業變得更加彈性。最後，分析結果用以最佳的資料視覺化方式做呈現，可以在不簡化資訊情況下，降低複雜資料的理解門檻，有助於院內臨床醫生一目了然資料分析的結果。
In recent years, hospital clinicians are facing immense challenges in deciding which antibiotic treatment to prescribe for a patient. The rise of antibiotic-resistant bacteria has led to an urgent need for rapid detection of drug resistance in clinical samples. Whole-genome sequencing (WGS) has been used to sequence and analyze the bacteria genomes in order to understand the resistant profile. However, analysis of WGS require integration of many tools and databases, which are not user-friendly. For this reason, we developed C?ab, a web server for bacterial genome analysis. C?ab supports automated genome assembly from third-generation sequencing data and invokes a series of sequential analysis stages. These stages include Identification of microbial strains, antibiotic-resistance detection, and virulence estimation pipelines. Users can access C?ab through web browsers to set parameters manually or automatic and execute pipelines, which have been implemented to run in non-concurrent on our local server cluster. C?ab provides interactive web interface and data visualization, which offers to hospital clinicians great flexibility for complex bacterial resistance and virulence data analysis
體型與犯罪的關係：以BMI為例;The Relationship between Body Type and Crime：Illustrated by the Case of Body Mass Index
No full text[[abstract]]本研究旨在透過犯罪生物學家Kretschmer 和Sheldon的觀點，檢證犯罪與體型之間的關係。體型分類以計算體型的身體質量指數（Body Mass Index）為基準，將體型分為過輕、正常、過重型，並針對台灣地區2005-2013年台灣21個縣市的犯罪資料，利用最小平方法與固定效果模型進行實證分析。 研究結果顯示，體型與犯罪間具有顯著關係，正常型的犯罪率最高，且在暴力、故意殺人、詐欺背信與竊盜發生都有顯著正相關；過輕型與犯罪率次高，與正常型的差別在於與暴力、故意殺人無相關；肥胖型的犯罪率最低，與犯罪都呈反向關係。進一步討論體型與經濟型犯罪之關係，結果顯示，兩者間並無明顯關係，但非經濟型犯罪在肥胖型呈現顯著負相關。據此可知，犯罪生物學家所指稱的體型學說，在現今影響犯罪的成因之中不盡完全適用。
This study aims to verify the relationship between body type and crime from the viewpoint of Kretschmer and Sheldon, who are devoted to researching biological criminology. The researcher uses Body Mass Index which is commonly used to classify underweight, normal weight and overweight instead of body type, and the cross-regional panel data of 21 countries during the period 2005-2013 in Taiwan. The empirical result from least squares method and fixed-effects model regression found that there are significant relationship between Body Mass Index and crime rate, normal weight has high crime rate and high-positive correlation with violent crime, murder, fraud and larceny, underweight has median crime rate and positive correlation with fraud and larceny, overweight has low crime rate and negative correlation with all. Take economic crime for further analysis, the results reveals body comrormation has no significant effect on economic crime but has negative correlation with non-economic crime. Therefore, body type theory is not fully applicable to the cause of crime now
區位模型與工資探討;A Study of Location Model and Wage
No full text[[abstract]]一間商店的員工，為生產要素供給者同時也是商品需求者，有沒有可能員工不支持自己商店的產品，反而跑到另一家商品類型相同的店家消費？本文利用Hotelling模型考慮員工同時也是消費者的角色，最後發現此模型下，員工並不會分離於兩家廠商消費和工作，並且在員工會在原本上班地點採買時，兩家廠商的區位決策會位於城市的兩個端點。而工資透過靜態分析結果：一間廠商工資的上升，將使這家廠商的區位決策選擇遠離另一家廠商，以避免自己在處於不利條件下，與他廠進行更激烈的價格競爭。並且即使原本兩家廠商名目工資一樣，若採取的員工折扣有所不同，將導致兩家廠商的員工所獲得的實質工資有所差異，進而讓交通成本與員工工作外部成本對廠商區位決策造成影響
食品安全訊息對臺灣肉類需求的影響;Effect of Food Safety Information on Taiwan’s Meat and Fish Demand
No full text[[abstract]]近年來隨著國人的所得提高，消費者開始注重食品的安全。而隨著資訊科技的發達，人們接收訊息的方式也越來越多元化。故本研究採用線性近似理想需求體系（Linear Approximate Almost Ideal Demand System, LA/AIDS）模型，並納入食品安全訊息指標於模型中，建立一肉類需求體系模型。資料研究期間為1985年到2015年的年資料。指標則利用臺灣1985年到2015年間，電視、報章雜誌、網路對於食品安全事件的報導數量建立三個不同來源的食品安全訊息指標。本研究除了計算豬肉、牛肉、魚肉、羊肉、家禽肉的需求彈性值外，也將探討食品安全訊息指標對於臺灣肉類需求的影響。實證結果顯示，不同來源、個別肉類的食品安全訊息指標對於該肉類皆無顯著的影響。五種肉類自身的網路食品安全訊息指標對自身肉類需求的影響最大
我國中央銀行貨幣政策法則之文獻回顧;The Literature Review of Monetary Policy Rule of the Central Bank of the Republic of China
No full text[[abstract]]因著貨幣政策的重要性，以及Taylor法則在描述美國貨幣政策反應函數的良好表現，在近二十餘年之間吸引了許多國內學者針對我國中央銀行的貨幣政策法則進行研究，但由於還沒有文獻將此一面向的發展進行系統性的整理，因此本研究以較為深入的介紹，將過往與台灣貨幣政策法則相關的論著進行回顧，並將這八篇期刊論文根據研究目的之演進來進行分類與比較，希望能夠提供給未來有志進行後續探討的研究者一些參考。回顧過往與台灣貨幣政策法則相關的論著，可以發現這些研究多以隔夜拆款利率為政策指標，並在考慮央行可能存在之不對稱反應下選擇門檻模型，以及針對變數間顯現之內生性而使用工具變數法；而實證的結果也顯示政策的施行具有反循環之特色，以及即時資料比修正資料更適合用來分析央行的施政行為。比較明顯的分歧則出現在以匯率為門檻變數的情況下，部分研究認為央行的反應符合其對外宣稱之維持動態穩定，但也有些研究則發現政策的施行會造成弱勢貨幣的結果
油品通路經營績效之研究;A study on operating performance of gasoline distribution channel.
No full text[[abstract]]油品通路市場加油站在自由化之後，從初期的通路爭奪戰，歷經高油價期間政府因應通貨膨脹之壓力，限制中油既有供油策略，取消各項優惠，及相關法令規章的改變，墊高加油站經營人力及營運成本，然後又因原油供需失衡，油價下跌所產生利潤縮減等，讓加油站業者經歷一次又一次重新洗牌的經營困境。本研究取自個案公司的油品銷售相關數據進行實證研究，利用SAS統計軟體為輔助工具，以相關性分析來檢定各變數之關聯，用迴歸分析探討自變數與經營績效的解釋能力，來找出加油站經營績效的決定因素，同時指出研究之限制，及對未來研究者與加油站管理者提出建議。研究結果顯示(1)在財務構面間接及直接行銷費用的投入僅部分支持與經營績效呈正相關之假說。(2)在顧客構面汽機車加油站來客數及汽柴油日均發油量均與經營績效呈顯著正相關。(3)在營業特性加油站營運年數與經營績效呈顯著正相關。(4)在站點區位加油站同馬路同方向競爭站點數與經營績效呈顯著負相關。(5)在加油站之站長特質站長年資與經營績效呈顯著正相關。關鍵字：油品通路、迴歸分析、經營績效
The operation of the gas station after the liberalization of the market from the initial channel battle, to the high oil prices suffered by the reduction during this period. And then changes in the relevant laws and regulations of the government to increase the manpower and operating costs of gas stations. Finally because of the imbalance between supply and demand of crude oil brought a fall and reduce the profits. So that the gas station operators experienced again and again to re-shuffle the operating difficulties.This study is taken from the T company's oil sales data for empirical research, we use the SAS statistical software as an auxiliary tool to test the correlation of variables by correlation analysis. We also use regression analysis to explore the explanatory power of the independent variables and management performance. The determinants of the performance of the gas station, the limitations of the study, and suggestions for future researchers and gas station managers.The results show that: (1) The indirect and direct marketing costs of investment only partially support the hypothesis that business performance is positively correlated in the financial aspect. (2) There is a significant positive correlation between the customer number and average daily fuel consumption in the customer aspect. (3) There is a significant positive correlation between the business years and the performance in business characteristics. (4) There is a significant negative correlation between the number of gas stations competing with the road in the site location and the performance of the stations. (5) The station master 's seniority has significant positive correlation with the operating performance in the gas station.Key word：gasoline distribution channel，regression analysis，the operating performanc
企業社會責任揭露與權益資金成本之關聯性;A Study on the Relationship between Corporate Social Responsibility and Cost of Equity Capital
No full text[[abstract]]近年因台灣發生一連串食安、工安及環境污染事件，政府開始積極推動企業社會責任的落實，以期望透過企業社會責任報告書的揭露，來提高企業的透明度與當責性，以降低企業管理當局與社會大眾、外部投資者等之間資訊不對稱的情形。本研究以2013至2016年之台灣上市櫃公司為研究對象，檢視企業社會責任的揭露與權益資金成本之關聯性，在控制了公司規模、負債比率、淨值市價比、法人持股比率、獲利能力、產業別及年度別等預期會影響權益資金成本的變數後，實證結果顯示，企業社會責任報告書之揭露及強制性揭露企業社會責任報告書與權益資金成本皆呈顯著負相關，且若報告書有經第三方確信會增強其效果，但自願揭露企業社會責任報告書會減弱與權益資金成本的負相關，若強制前已自願揭露企業社會責任報告書會增強與權益資金成本之負相關。強制性揭露企業社會責任報告書對權益資金成本之影響效果大於自願揭露的效果。被政府強制規範揭露且強制後第一次揭露企業社會責任報告書之公司，其強制前後的權益資金成本沒有顯著的變化。
In recent years, a series of food, safety and environmental pollution incidents have occurred in Taiwan. The government began to actively promote the implementation of corporate social responsibility. In order to promote the disclosure of corporate social responsibility report to improve the transparency and accountability of enterprises to reduce the enterprise management and the public, external investors and other information asymmetry between the situations.This study focuses on enterprises registered in the public stocking market between 2013 and 2016 in Taiwan. In order to examined the relationship between the disclosure of corporate social responsibility and the cost of equity capital. In the control of the company's size, debt ratio, book to market ratio, institutional investors holding rate, profitability, industry, yearly. The empirical results indicated that there is a significant negative correlation between the disclosure of corporate social responsibility report and the mandatory disclosure of corporate social responsibility report and the cost of equity capital. The corporate social responsibility report has been assured by a third party will enhance its effectiveness, but voluntarily exposing the corporate social responsibility report will be less negative correlation with the cost of equity capital. If voluntary disclosure of corporate social responsibility reports before mandatory will increase the negative correlation with equity capital costs. Mandatory disclosure of corporate social responsibility report on the impact of the cost of equity capital effect is greater than the effect of voluntary disclosure. The corporate social responsibility report disclosure had been mandated and for the first time, there is no significant change in the cost of equity capital before and after enforcement