Course background and introduction¶
from IPython.display import IFrame
IFrame(src="https://cdnapisec.kaltura.com/p/2356971/sp/235697100/embedIframeJs/uiconf_id/41416911/partner_id/2356971?iframeembed=true&playerId=kaltura_player&entry_id=1_qa61sa8g&flashvars[streamerType]=auto&flashvars[localizationCode]=en&flashvars[leadWithHTML5]=true&flashvars[sideBarContainer.plugin]=true&flashvars[sideBarContainer.position]=left&flashvars[sideBarContainer.clickToClose]=true&flashvars[chapters.plugin]=true&flashvars[chapters.layout]=vertical&flashvars[chapters.thumbnailRotator]=false&flashvars[streamSelector.plugin]=true&flashvars[EmbedPlayer.SpinnerTarget]=videoHolder&flashvars[dualScreen.plugin]=true&flashvars[hotspots.plugin]=1&flashvars[Kaltura.addCrossoriginToIframe]=true&&wid=1_xeoqsxwa" ,width='800', height='500')
Please access the video via one of the following two ways
Contents¶
Introduction to Machine Learning and Its Successes
Deep Learning, Deep Concern
A First Glance to Mathematics of Deep Learning
What Will We Study in This Course
Four major methods of scientific research¶
Experimental Science
Theoretical Science
Computational Science
Data Science
The era of the artificial intelligence¶
Success Stories - I: Image Classification¶
In 1998, LeCun etc. proposed the neural network LeNet-5 based on the convolution and applied to the hand-written digits recognition successfully. Hence LeNet-5 is also called the first convolutional neural network(CNN) successfully applied.
In 2012, Hinton and his student Alex joined the graph recognition competition by ImageNet and improved the accuracy significantly with the CNN AlexNet.
In 2013 Google purchased a Canadian startup company on neural network, DNNResearch. It’s set up by Geoffrey Hinton and his graduate students Alex Krizhevsky and Ilya Sutskever in 2012.
In 2015, Kaiming He etc. proposed the ResNet CNN structure, which has become a state-of-art CNN structure and widely used in industry and discussed in academia.
Afterwards CNN is widely used in the field of computer vision and breaks the records ceaselessly.
Success Stories - II: AlphaGo & AlphaZero¶
From 2016 to 2017, the AI program AlphaGo developed by Google DeepMind beat down all the Go champions worldwide.
2018 AlphaZero gave a unified principle for many other board games
The CEO of Google DeepMind Demis Hassabis announced to integrate AlphaGo with medical, robots and so on. They can learn by themselves since they are artificial intelligence, and transfer learning can be done with enough data.
Success Stories - III: Auto-driving¶
In 2009, Google proposed the plan to replace human driving with softwares. Afterwards, many large technology companies such as Tesla, Google, Uber and Benz devoted a lot on investigating the technology of automated driving.
There are about dozens of companies focus on automatic drive techniques from L2 to L4.
In many countries, road examination autonomous vehicles are allowed with applications.
Tesla – Autopilot; Google – Waymo; Baidu – Apollo; GM – Cruise; Volkswagen – DAS Autonomy…