AdSense

Thursday, April 27, 2017

AI

Chapter 1 Why AI now?

Three types of AI:
(1) rule-based
(2) probability/statistics-based in Bayesian network 
(3) deep learning in neural network

Current web search by Google is based on the page rank algorithm, which highly valuate a website with many links from others; many links mean that the web site is more trustworthy.

The next phases of web search are:
(1) semantic search by Google, which understands what people really want, not just search based only on a keyword used
(2) Siri, an intelligent personal assistant and knowledge navigator, by Apple
(3) graph search by FaceBook - e.g., "What is my friend's favorite song?"
Each player tries to provide necessary info after listening to what a user says; i.e., each company wants to dominate a gateway to the mobile internet.

Big data consists of:
(1) structured data (e.g., customer name, sales figures)
(2) unstructured data (irregular data like inquiries to a call center)


Chapter 2 What is intelligence?

Artificial Intelligence (AI): technology that substitutes humans' intelligent activities with machines such as computers and softwares on it


Chapter 3 Value of Intelligence

Requirements for a project:
(1) What is it for? The more a goal is concrete, the higher the probability of success is.
(2) Feasibility. Is technology developed enough to make it happen? How about human resources?
(3) Is it meaningful for a society?
(4) Why now? Is it timely?
(5) Evolvability / expansibility. What kind of technology is derived from it?
(6) Publicity. Is it interesting for a society?
(7) Motivation. Does a project motivate people?


Chapter 4 Issues of AI

(1) Risks of when people are too dependent on machines / systems.
(2) Concerns for people's employment and existence value replaced by AI.

(1) Risks of when people are too dependent on machines / systems.

Automated driving
Flash crash in stock markets

(2) Concerns for people's employment and existence value replaced by AI.


AI for chess,etc.
(1) game tree exploration (forecasting)
(2) current status / condition analysis by evaluation function
Also, AI does not get tired, feel a fear or sleepy, and lose concentration.


No comments:

Post a Comment

Deep Learning (Regression, Multiple Features/Explanatory Variables, Supervised Learning): Impelementation and Showing Biases and Weights

Deep Learning (Regression, Multiple Features/Explanatory Variables, Supervised Learning): Impelementation and Showing Biases and Weights ...