Hi Emahn,

Standard machine learning techniques like multiple linear regression 
(http://en.wikipedia.org/wiki/Linear_regression) should give reasonable results 
on your application, which you could then use as a baseline against which to 
evaluate more sophisticated (non-linear) methods like decision trees or deep 
learning. 
 
For a basic grounding in practical ML, I would recommend Machine Learning: A 
Probabilistic Perspective by Kevin Murphy, or Pattern Recognition and Machine 
Learning by Christopher Bishop. 

Ricky

 
Date: Fri, 19 Sep 2014 21:12:55 -0700
Subject: [agi] UPDATE: after 1 year. Please advise..again :]
From: [email protected]
To: [email protected]

Dear Singularity and AGI readers, 
I'd like to give you guys an update on what's happened since our last round of 
emails, and to ask for some help (more involved this time).

You may remember me from an email thread I started about a year ago 
entitled,"Student here, please advise."link to the archive: 
https://www.mail-archive.com/[email protected]/msg04433.html

At that time, I was asking for advice on which university major to pursue, and 
what I can do to help further the advancement of AI. I received a surprising 
amount of helpful suggestions from many members of this community, including 
Jan Klauck, Ben Goertzel, Mike Archbold, Matt Mahoney, Sergio Donal, and 
Anastasios Tsiolakidis.

I wrote this email in part to thank all of you for giving me the guidance and 
inspiration to follow through with what I dreamed of doing ever since I read 
the Cosmist Manifesto. 
Ever since then, I've worked a lot on improving my programming skills. I've 
learned Java, Python, Ruby, Ruby on Rails, Javascript/jQuery, HTML/CSS, and C++.
This summer I secured a job as the lead developer at a small start up company - 
focused on developing a form of narrow AI - with the purpose of helping 
employers sort and rank resumes. 
The software we developed reads through resumes and ranks the applicants based 
on individual criteria such as business, software, and marketing experience. We 
released the beta version mid August (a few weeks ago), and one corporation has 
already signed on with us!
Here's where I need your help:
The way our algorithm works is admittedly pathetic ... we feed it a set of 
hundreds of terms and phrases relevant to a particular criteria, and after some 
basic word fuzzing, the program searches through the resumes, and awards points 
based on the occurrence of matching terms. It then adjusts the overall ranking 
of applicants using "weights" assigned by the recruiter with a slider for each 
criteria. The whole application is written in Ruby within the Rails framework 
and hosted on heroku. 
Here's what part of the UI looks like:

I'm looking for some guidance on how to implement actual AI principles like 
neural networks to improve the ranking process. I have a lot left to learn as 
far as programming anything resembling AI goes, and I know there is a wide 
breadth of information on the topic. But the reason I'm coming to you guys 
first before diving into the topic literature on my own is to hopefully get 
some pointers on what concepts, methods, and/or implementations of AI to focus 
on. 
I would really appreciate any specific advice on how to start implementing the 
exact type of AI I'm looking for. What set of concepts and tools in the AI 
toolkit do you guys think would be most relevant to the problem I'm trying to 
solve? Any help will be greatly appreciated.

That's all for now... and as for my major, I finally chose to go with Cognitive 
Science with a focus in computation. I'll be transferring to a UC(University of 
California) next fall, hopefully it'll be UCLA. 
Thanks again for the help fellow AGI'ers. I hope we can make a lot of progress 
toward AGI in the near future and maybe even get close to singularity!! :D 
Needless to say, I'm super excited to be a part of it all. 

Regards,
Emahn








  
    
      
      AGI | Archives

 | Modify
 Your Subscription


      
    
  

                                          


-------------------------------------------
AGI
Archives: https://www.listbox.com/member/archive/303/=now
RSS Feed: https://www.listbox.com/member/archive/rss/303/21088071-f452e424
Modify Your Subscription: 
https://www.listbox.com/member/?member_id=21088071&id_secret=21088071-58d57657
Powered by Listbox: http://www.listbox.com

Reply via email to