[Pycon] [new paper] "Harshinee Sriram" - The art of building a sophisticated psychologist with Python

info a pycon.it info a pycon.it
Mar 4 Dic 2018 00:58:05 CET


Title: The art of building a sophisticated psychologist with Python
Duration: 45 (includes Q&A)
Q&A Session: 15
Language: en
Type: Talk

Abstract: > "Happiness comes in waves, and it will find you again".

Why Mental Health?
------------------
According to the statistical analysis by the World Health Organization (WHO), a suicide occurs every 40 seconds. Drug addiction alone has claimed over 130,000 people (US) in the past 3 years, according to the National Institute on Drug Abuse.

What does this talk contain?
----------------------------
The human mind has always been one of the most complicated subject areas to completely understand, and new discoveries are made every day.  So, is it possible to create a fully functioning, efficient, and coherent machine that can act as a fully capable psychologist which can be mass produced in order to curb the existing deficiencies that exist in the areas of the numbers and effectiveness of current mental health setups? 

This talk covers everything from building the first neural network (multi-layer perceptron MLP) and understanding its processing to building one that understands and analyzes human tone and emotions. It then covers the current research done in the field of human psychology for robotics and AI, and also presents examples of currently existing models - including what can be improved. This talk also presents the current challenges and what could possibly be done to eliminate them, starting from agent architectures, to therapeutic games, and ultimately to human behaviour. Moreover, it inspires the listeners to delve into creating solutions for conditions that could potentially help millions.

Prerequisite(s)
---------------
Calculus (especially knowledge about derivatives)



Tags: [u'computer-science', u'healthcare', u'pyladies', u'python3', u'Artificial Intelligence', u'Algorithms', u'robotics']


Maggiori informazioni sulla lista Pycon