Machine Learning and AI by Dr. Gang li, Deakin University, Australia.
Machine Learning and AI by Dr. Gang li,
Deakin University, Australia.
10-06-2020 from 03:00 P.M to 04:20 P.M.
Notes prepared by Dr. Phrangboklang Lyngton Thangkhiew,
Assistant Professor Senior Grade - I,
School of Computer Science and Engineering (SCOPE),
VIT University, Chennai.
Talk on- When Artificial Intelligence Meets Big Data?
Some Research Theme of Dr. Gang Li
Periodic Behaviour- Regular or Normal Pattern
Information Abuse Prevention- Identification of strength or weakness of an individual
Business Intelligence Applications- Come up with new research from current business.
Intelligence Application
Create applications that mimic human thinking. E.g. IBM Watson, Stanley (Auto Driving Car).
It started as earlier as 1980s.
The computer can play chess.
Deep Blue vs. Kasporov. Deep Blue wins
Moore’s Law
Every year the number of transistors on an IC will be double and so does the storage.
This provides us more capacity to store and collect data.
Big Data Age
With colossal data, it becomes difficult to handle more data.
Difficult to store or processing.
From computer science, one solution is to horizontal scaling.
Horizontal Scaling- Group of Computers work together to store or process together. This is known as Cloud Computing.
4th Paradigm of Science
1000 years ago: Empirical + Experimental
Theoretical
Computational
Data Exploration
Collect Data of enormous size
Not easy to process with easy methods
Call as science or 4th Paradigm of Science
CSIR in Bangalore is one such institution in India to do such research
Why AI?
Data Processing/Analysis has become the core of CS.
When massive data involve, it becomes difficult to process with traditional methods.
There is a need for advance and proper methods to process such Data.
What is Machine Learning?
Machine Learning can be used in.
Face Recognition
Human thinking like
Automated Mail Activity
Bio Informatics
Business informatics
Machine Learning Subclass.
Supervised learning- Human help computer to learn
Unsupervised learning- Train the computer with lots of data
Structured Learning-
Input and output are of different format
Reinforcement Learning- Group of annotation to help computer maneuvers into a near-optimal solution
Deep Learning
Changing the world in so many aspects.
Theme 1: Behaviours Informatics
Tourist Management Analysis
Traditional methods are not suitable for the survey. E.g., Participants cannot remember the place, name, etc.
Tourist place many photos online. With such photos, we have information such as
Time
Location
With such photos, tourist hot spots (Area or Route) can be identified by using Behaviours Informatics.
It can be used to predict behavior for instance.
Customer details can be collected from their online activity. For example, the picture below shows particular information about a tourist.
From this GPS location, we can know how the customer moves around.
Periodic Behaviour Mining.
Application of Periodic Behaviour.
Recommends customer to buy essential things based on the data collected
Problem on Data Acquisition.
Privacy- For e.g., photo can be without GPS tag.
Apriori Algorithm- Prof Agarwal
Example: Can be used to find popular items together, such as.
Things that people tend to buy together in a shopping mall.
Or people that travel together.
Can also be used to distinguish the relationship.
Some Research perspective that can impact in 2025
Privacy Preservation.
Topologic Data Analysis.
Summary
Today the world's is the era of big data.
The idea is to find a smart method to collect data.
And to use AI and ML to do research
Privacy must be respected in doing such research.
Topological research is very promising.
**************************
Comments
Post a Comment