產業創新應用
Innovations and Applications
▎AI新聞可信度辨識系統
AI News Credibility Identification System
►國立臺灣師範大學 王維菁教授 / Prof. Wei-Ching Wang, National Taiwan Normal University
為了因應假新聞的氾濫與傳佈，本團隊將研究透過人工智慧(Artificial Intelligence, AI)作為偵測假新聞的工具。從語言特徵中我們使用自然語言處理NLP的文本分析，結合中研院CKIP中文斷詞系統與LSTM以及Bert的深度學習技術，找出本土化的AI假新聞查核技術發展方向，設計出一套辨識假新聞的系統模型，讓產官學界在假新聞防治上有不一樣的思考方向，並當作對於大眾的數位素養教育可運用的工具，或與各大社群網站平台合作，提供遏止假新聞之有效的工具。
In response to the proliferation and spread of fake news, we use Artificial Intelligence (AI) as a tool to detect fake news. We apply natural language processing (NLP) to processing text analysis through Traditional Chinese Language
features, combining the Academia Sinica CKIP Chinese word segmentation system with the deep learning technology of LSTM and Bert. We are eager to find the development direction of localized AI fake news checking technology and design a system model for identifying fake news. With our research, we try to provide the industry, government, and academia with a different way of dealing with fake news. Also, our research will be helpful as a tool to promote digital literacy education, cooperate with social network platforms, and curb fake news.
Reference :
- AI新聞可信度辨識系統網址：http://140.122.86.116/fnds.php
▲ AI新聞可信度辨識系統的教育應用
Educational application of AI news credibility identification system
此研究歸屬科技部 AI 專案計畫執行成果，詳細資訊請參考附錄之計畫總表第 22 項
For the name of the project which output this research, please refer to project serial no. 22 on the List of MOST AI projects on Appendix
▎自然廣告圖片生成
Native Image Ads Generation
►國立清華大學 吳尚鴻教授 / Prof. Shan-Hung Wu, National Tsing Hua University
藉由我們提供的自然圖片生成技術，公司可以找到最符合產品形象的圖片放進相關的廣告文案中。由於此項技術的圖片生成過程皆是由深度學習網路來處理，過程中不包含任何版權圖片的參照，所以也可以避免實務上經常遇到的版權問題。
我們和AppFinca 公司合作，透過此項技術協助其手機應用程式「Flora」推廣廣告，並且成功將其推上台灣區Apple App Store免費生產力工具分類排名第一。
With native image ads generation, companies are able to generate
images that are most suitable for their products through our technique and apply them on ads without copyright concern.
We have cooperated with AppFinca Inc. and helped their mobile application “Flora” to reach number one on the category of productivity tools on Apple App Store of Taiwan.
Reference :
▲ 自然廣告圖片生成之輸入與輸出。
The Input/Output of Native Image Ads Generation process.
▲ 吳尚鴻教授與 AppFinca 副總一同與科技部記者會。
Prof. Shan-Hung Wu and the vice president of AppFinca Inc. co-attended on the press conference of MOST.
此研究歸屬科技部 AI 專案計畫執行成果，詳細資訊請參考附錄之計畫總表第 6 項
For the name of the project which output this research, please refer to project serial no. 6 on the List of MOST AI projects on Appendix
▎腦影像之精神疾病輔助診斷平台
Al-Based Web Diagnostic System for Phenotyping Psychiatric Disorders
►臺北榮民總醫院 蔡世仁教授 / Prof. Shih-Jen Tsai, Taipei Veterans General Hospital
相對於國際上目前類似產品準確率約為71%-76%，本團隊所發展之思覺失調症腦影像診斷平台，可達91.7% 之準確率且提供精準細節，包含腦病變位置、可能患病機率、分類器判定是否患病，並呈現可能異常區域名稱與座標。如此細節都將有助於臨床醫師結合其學識與專業判斷、使得非專業人員正確認識該疾病，並有效遏止潛在利用精神疾病規避責任者。此腦影像平台的出現，不僅為臨床醫師提供輔助診斷，更將真正成為精神疾病診斷之客觀衡量指標。
Mental illness is a critical health issue in Taiwan. Psychiatric diagnosis is largely based on self-reported or symptomatic criteria without objective findings. This state-of-the-art psychiatric diagnostic platform using standardized brain imaging data and improves the efficiency and accuracy of psychiatric diagnosis. Our platform can easily identify the deficiency in brain regions associated with schizophrenia, provides a novel way to evaluate mental illness and its progression, with powerful visualization.
This evidence-based, AI-assisted psychiatric diagnosis platform, validated with large-scale standardized brain imaging. We believe this state-of-the-art psychiatric diagnostic tool can help professionals assess disease progression and treatment outcomes almost in real time. Most importantly, this will bring the patients, caregivers, and general population a new perspective on the nature of the mental illness, significantly reduce the cost of healthcare insurance and socioeconomic costs for mental disorders, and promote mental health in the general population.
Reference :
◀ 腦影像之精神疾病輔助診斷平台
Al-Based Web Diagnostic System for Phenotyping Psychiatric Disorders
◀ 本技術已榮獲第十七屆國家新創獎之殊榮
This AI-Based Web Diagnostic System Granted the 17th National Innovation Award in 2020
此研究歸屬科技部 AI 專案計畫執行成果，詳細資訊請參考附錄之計畫總表第 30 項
For the name of the project which output this research, please refer to project serial no. 30 on the List of MOST AI projects on Appendix
▎獲得2020國家新創獎
Won the National Innovation Award, 2020
►國立臺灣大學 黃升龍特聘教授 / Distinguished Prof. Sheng-Lung Huang, National Taiwan University
本團隊技轉安盟生技(AMO)所開發之台灣原創的高解析活體光學影像系統(ApolloVue® S100)，在國際間具有最高的三維解析度，可即時呈現人體皮膚之完整表皮層及上真皮層結構，結合智能影像導引快速切換橫切面或縱切面影像模式，可以更有效率地探索患部皮下立體資訊，減少遺漏病變檢查的機會，具高度臨床應用的潛力。目前ApolloVue® S100已獲美國FDA以及台灣FDA認證，預期商機市場大。本團隊持續與AMO合作導入AI，以具細胞解析度影像應用在皮膚疾病的診斷與治療，並已開展國際臨床試驗。
The team has transferred technology to Apollo Medical Optics (AMO) to develop a high- resolution in vivo optical coherence tomography system (ApolloVue® S100), which has the highest three-dimensional resolution in the
world and can instantly display the complete epidermis and upper dermal structure of human skin. Integrated with intelligent image guidance to quickly switch between the cross-sectional and en face image modes, the system can efficiently explore the 3D information of the lesion for clinical applications. With the FDA 510(k) notification, AMO is ready to realize the market potential and carry out international clinical trials. At present, the continuing collaboration between the team and AMO on AI algorithms is successful in the diagnosis and treatment of skin diseases with cell-resolution images.
▲ 賴副總統頒授國家新創獎
▲ 安盟生技團隊
此研究歸屬科技部 AI 專案計畫執行成果,詳細資訊請參考附錄之計畫總表第 27 項
For the name of the project which output this research, please refer to project serial no. 27 on the List of MOST AI projects on Appendix
▎影像辨識應用於長照機構
Image recognition used in long-term care institutions
►國立清華大學 孫民副教授 / Associate Prof. Min Sun, National Tsing Hua University
CarePLUS.ai是透過影像辨識技術來偵測年長者的居家活動狀況，並在意外發生時通知照護者。
目前已實際建置CarePLUS.ai系統至使用者家中與長照機構，此專案亮點為：
- 居家用戶的CarePLUS.ai系統運行已超過六個月。
- 進入長照機構實際測試後，將進入商用階段。
CarePLUS.ai uses image recognition technology to detect the home activities of the elderly and notify the caregiver when an accident occurs.
At present, the CarePLUS.ai system has been actually built into users’ homes and long-term care institutions. The highlights of this project are:
- The CarePLUS.ai system for home users has been running for more than six months.
- After entering the actual test of the long-term care institution, it will enter the commercial stage.
Reference :
此研究歸屬科技部 AI 專案計畫執行成果,詳細資訊請參考附錄之計畫總表第 12 項
For the name of the project which output this research, please refer to project serial no. 12 on the List of MOST AI projects on Appendix
