Profile

Cover photo
Phillip Adkins
Works at Data Analysis and Machine Learning Engineer
Attended University of Southern California
Lives in Silicon Valley
569 followers|155,748 views
AboutPostsPhotosVideos

Stream

Phillip Adkins

Puzzles/Riddles  - 
 
Imagine i have a target binary random variable, and "evidence" of the target value.

The evidence is of the form P(y=1). However, the evidence can be "corrupted" - in a way that can arbitrarily lower the reported value of P(y=1) from its true value; so a P of zero doesn't mean that y=0, while a P of 1 DOES mean that y=1.

I can have anywhere from 1 to an infinite pieces of "evidence" for each sample - this number being randomly selected for each sample.

What is the best way to combine this evidence for an overall probability score?

2
Iwuchukwu Obieze's profile photoJoe Schaefer's profile photoPhillip Adkins's profile photo
8 comments
 
+Joe Schaefer yep - probabilities between 0 and 1
Add a comment...
 
Smart coders wanted for short, intermittent, one-off projects.

Think "i need you to write me a few classes to do task X for $1000".

I want several smart, reliable guys with clean code and great communication skills on-call.

Respond here or PM me with email address.

Thanks ;)
4
Mel Pieceofchalk's profile photoPhillip Adkins's profile photo
6 comments
 
+Mel Pieceofchalk

Are you signed up with coder's clan?

Can you send me a resume? 
Add a comment...
 
A buddy of mine is making some pickups for me. They are SO SWEET looking, and hand-wired to my specs. They can make anything you can dream up no matter how extreme. Check out it out. WIP, but I'm so excited:

I told him I'd spread the word, so:

http://facebook.com/protoguitars
11
Add a comment...

Phillip Adkins

Shared publicly  - 
 
I'm terrible at ties. It is my goal to now become tie ninja.
 
A Demonstration of How to Quickly Tie a Tie ‘Like a Ninja’
Taras Kulakov the Crazy Russian Hacker shows how to tie a tie "like a ninja" in a demonstrational video. The resulting knot appears to be a Nicky knot, and although traditionally ninjas were not kn...
1 comment on original post
3
Phillip Adkins's profile photoVeronika Franz's profile photo
12 comments
 
that's ok... I understand that. 
Add a comment...

Phillip Adkins

Shared publicly  - 
 
Interesting talk, an analysis which I haven't heard before
 
Good talk on the basic properties of deep learning: Signal and Image Classification - Stephane Mallat Technion lecture
The interesting message is that you would not easily get the results of Deep Networks by approaching the problem theoretically only. These results are 10+ years old and have been obtained by studying neuroscience, and other disciplines, and applying the most promising and evident models.
View original post
2
1
Add a comment...

Phillip Adkins

Puzzles/Riddles  - 
 
Find the next number in the sequence:

9, 8, 6, 6, 5, 5, 1, ?
4
Andres Rivera-Lasalde's profile photoCarlos Eduardo Olivieri's profile photoPhillip Adkins's profile photoNoel Artiles-Leon's profile photo
42 comments
 
+Andres Rivera-Lasalde Excellent observation and question! ... Especially now that the acceptable secuences are no longer limited to integer secuences; one must be able to find many other way to continue the inicial sequence {9, 8, 6, 6, 5, 5, 1, ...}
Add a comment...

Phillip Adkins

Shared publicly  - 
 
Supah Sweet
 
toolz: A functional standard library for Python.
View original post
1
Add a comment...
 
The default behavior of pandas to_csv is to add a new index - a brand new column of on-the-spot generated indices - to the csv.

You can deactivate this by setting index=False

However, I believe it should be False by default; why should an alteration to the structure of your data be the "default" option?

You might as well also append a time-series clip of the neigh of a shetland pony by default.

Also, with the default settings - if you repeatedly load and save a csv, it will add column after column of index.

This is obviously not ideal behavior.

index=False by default - who's with me?!
2
Add a comment...

Phillip Adkins

Discussion  - 
 
Why does this group exist when there's a much larger (125k users) group also called "Python"?

I vote that we destroy this group and consolidate under the other to strengthen it.
13
Phillip Adkins's profile photoJean-Michel Rubillon's profile photoTom Hallam's profile photo
17 comments
 
A quick search of communities gives 8 other groups (additional to the first 2) that are also just called "Python" 
Add a comment...

Phillip Adkins

Discussion  - 
 
Python 3.5 to have "@" operator for matrix multiplication.

I hate it. Why not "x"?? 
The official home of the Python Programming Language
13
2
Bryan Bonvallet's profile photoHannes Ovrén's profile photoAnders Steen Christensen's profile photoPhillip Adkins's profile photo
14 comments
 
+Anders Steen Christensen

I literally hate the use of the shorthand character "@" to represent matrix multiplication in Python. Yes.
Add a comment...

Phillip Adkins

Shared publicly  - 
 
 
Hedgehog Inadvertently Plays a Respectable Measure of Jazz Just by Walking Atop Piano Keys
Marutaro, the adorable hedgehog who poses for silly pictures and loves to be tickled, inadvertently plays a respectable measure of jazz whenever he walks on the keys of the family piano.
View original post
1
Add a comment...

Phillip Adkins

Shared publicly  - 
 
Too cute to not share
 
Tiny Baby Hedgehog Tries To Mouth A Human’s Finger
A tiny 13-day old hedgehog tries very hard to mouth a human's finger without success in this absolutely adorable video posted by donburiko. via Say OMG
2 comments on original post
6
Add a comment...
Education
  • University of Southern California
    Physics, 2013 - 2013
  • Physics
Basic Information
Gender
Male
Work
Employment
  • Data Analysis and Machine Learning Engineer
    2011 - present
Places
Map of the places this user has livedMap of the places this user has livedMap of the places this user has lived
Currently
Silicon Valley
Links