时间：2022年5月25日（星期三）15:30-16:30
地点：腾讯会议：678-648-025
主题：认识人工智能中机器学习与深度学习（Machine learning and deep learning in artificialintelligence）
主讲人：孟飞（英国beat365官方网站入口）
简介：孟飞，副教授、硕士生导师。主要从事复杂系统建模与控制技术、人工智能理论及其应用方面的研究。在IEEE/ASME Transactionson Mechatronics、IEEE Transactions onVehicular Technology等学术期刊和国际学术会议上发表论文40余篇。主持及参与国家自然科学基金项目、上海市科学技术委员会科技等省部级项目8项。获得发明专利5项，实用新型专利5项。担任Advanced in mechanicalengineering、Sensors期刊的客座编辑，并同时为IEEE/ASME Transactions onMechatronics、IEEE Transactions onIndustrial Electronics、IEEE Transactions onControl Systems Technology等10余个SCI期刊审稿人。2015-2017年获Mechanical Systems and SignalProcessing杰出审稿人，2016-2021年国家自然科学基金通讯评议人。美国电气与电子工程师学会（IEEE）会员，美国汽车工程师学会新能源车辆委员会（SAE）委员。
摘要：近年来，以机器学习、知识图谱为代表的人工智能技术逐渐变得普及。从车牌识别、人脸识别、语音识别、智能助手、推荐系统到自动驾驶，人们在日常生活中都可能有意无意地用到了人工智能技术。最近几年，得益于数据的增多、计算能力的增强、学习算法的成熟以及应用场景的丰富，越来越多的人开始关注这个“崭新”的研究领域：深度学习。深度学习以神经网络为主要模型，一开始用来解决机器学习中的表示学习问题。但是由于其强大的能力，深度学习越来越多地用来解决一些通用人工智能问题，比如推理、决策等。本报告主要介绍人工智能、机器学习和深度学习的基础概念及相互关系。讲述三种主要的神经网络模型：前馈神经网络、卷积神经网络和循环神经网络及其在各领域的应用。
Artificial intelligence technologies represented bymachine learning and knowledge graphs have gradually become popular. Fromlicense plate recognition, face recognition, voice recognition, intelligentassistants, recommendation systems to autonomous driving, people may useartificial intelligence technology consciously or unintentionally in theirdaily lives. In recent years, thanks to the increase of data, the enhancementof computing power, the maturity of learning algorithms and the richness ofapplication scenarios, more and more people have begun to pay attention to thisnew research field: deep learning. Deep learning uses neuralnetworks as the main model and was originally used to solve the problem ofrepresentation learning in machine learning. But due to its powerfulcapabilities, deep learning is increasingly used to solve some generalartificial intelligence problems, such as reasoning, decision-making, etc. Thisreport mainly introduces the basic concepts and interrelationships ofartificial intelligence, machine learning and deep learning. Introduces thethree main neural network models: feedforward neural network, convolutionalneural network and recurrent neural network and their applications in variousfields.
