SERVING THE QUANTITATIVE FINANCE COMMUNITY

 
User avatar
Gamal
Topic Author
Posts: 2077
Joined: February 26th, 2004, 8:41 am

Quantitative methods in magic

March 6th, 2018, 11:28 am

Applications of maths and computers in magic are still undervalued. To start with https://arxiv.org/abs/1709.03803
 
User avatar
ISayMoo
Posts: 934
Joined: September 30th, 2015, 8:30 pm

Re: Quantitative methods in magic

March 8th, 2018, 10:33 pm

I've seen this paper. One of the worst ever.
 
User avatar
tagoma
Posts: 18233
Joined: February 21st, 2010, 12:58 pm

Re: Quantitative methods in magic

March 9th, 2018, 8:22 am

11 authors, several universities involved, and this research was funded.
 
User avatar
Gamal
Topic Author
Posts: 2077
Joined: February 26th, 2004, 8:41 am

Re: Quantitative methods in magic

March 9th, 2018, 8:41 am

Magic
 
User avatar
Cuchulainn
Posts: 56679
Joined: July 16th, 2004, 7:38 am
Location: Amsterdam
Contact:

Re: Quantitative methods in magic

March 9th, 2018, 11:24 am

N(N-1)/2 communication channels!

N = 11 => 55. Who are the silent partners?


Image
 
User avatar
ISayMoo
Posts: 934
Joined: September 30th, 2015, 8:30 pm

Re: Quantitative methods in magic

March 11th, 2018, 12:17 am

tagoma wrote:
11 authors, several universities involved, and this research was funded.

I've recently listened to a talk by a post-doc from Queen Mary University in London. She was optimising the random seed of the PRNG used by her model as a hyper-parameter, showing plots of reward function for "good seeds" and "bad seeds". To her credit, she did say "some people may find it controversial".
 
User avatar
Traden4Alpha
Posts: 23951
Joined: September 20th, 2002, 8:30 pm

Re: Quantitative methods in magic

March 11th, 2018, 12:24 am

ISayMoo wrote:
tagoma wrote:
11 authors, several universities involved, and this research was funded.

I've recently listened to a talk by a post-doc from Queen Mary University in London. She was optimising the random seed of the PRNG used by her model as a hyper-parameter, showing plots of reward function for "good seeds" and "bad seeds". To her credit, she did say "some people may find it controversial".

LOL!

Just wait until some AI finds the "good seed" that exactly outputs the "random" movements of the stock market.

Douglas Adams knew the Earth was just a giant computer. Ergo the market is really a PRNG.
 
User avatar
ISayMoo
Posts: 934
Joined: September 30th, 2015, 8:30 pm

Re: Quantitative methods in magic

March 11th, 2018, 12:27 am

I briefly wondered once if various Deep Learning models which require random inputs (e.g. autoencoders) could learn to predict the next random value generated, thus "cheating". But I doubt it's possible without a humongous training sample and large model capacity.
 
User avatar
Traden4Alpha
Posts: 23951
Joined: September 20th, 2002, 8:30 pm

Re: Quantitative methods in magic

March 11th, 2018, 1:02 am

Yes, the bit depth of the hidden state of the better PRNGs does seem to imply a humongous training sample.

However, for many numerical PRNG applications, even a volatile estimate of MSBs would be quite meaningful.

For cryptocurrencies, anything that helps predict outputs overlapping with the nonce would boost mining efficiency.

And for cryptography, perhaps sparse bit prediction would enable enough extraction of meaning to decode low entropy plain text.

(The deeper issue is: if Deep Learning models reach this level of sophistication, will they have the sophistication to withhold information on their results???)
 
User avatar
SWilson
Posts: 12
Joined: February 13th, 2018, 5:27 pm

Re: Quantitative methods in magic

March 12th, 2018, 10:30 pm

What about amplitude and vector quantization......
 
User avatar
Cuchulainn
Posts: 56679
Joined: July 16th, 2004, 7:38 am
Location: Amsterdam
Contact:

Re: Quantitative methods in magic

March 12th, 2018, 11:07 pm

ISayMoo wrote:
tagoma wrote:
11 authors, several universities involved, and this research was funded.

I've recently listened to a talk by a post-doc from Queen Mary University in London. She was optimising the random seed of the PRNG used by her model as a hyper-parameter, showing plots of reward function for "good seeds" and "bad seeds". To her credit, she did say "some people may find it controversial".

Image
 
User avatar
Traden4Alpha
Posts: 23951
Joined: September 20th, 2002, 8:30 pm

Re: Quantitative methods in magic

March 12th, 2018, 11:20 pm

Cuchulainn wrote:
ISayMoo wrote:
tagoma wrote:
11 authors, several universities involved, and this research was funded.

I've recently listened to a talk by a post-doc from Queen Mary University in London. She was optimising the random seed of the PRNG used by her model as a hyper-parameter, showing plots of reward function for "good seeds" and "bad seeds". To her credit, she did say "some people may find it controversial".

Image

Falsehoods also pass through the same three stages. Some even come out of stage 3 stronger than ever.
 
User avatar
mtsm
Posts: 340
Joined: July 28th, 2010, 1:40 pm

Re: Quantitative methods in magic

March 13th, 2018, 12:54 am

Gamal wrote:
Applications of maths and computers in magic are still undervalued. To start with https://arxiv.org/abs/1709.03803

Why do you think it's so bad? do you mean the execution of the authors is bad or do you mean the whole approach is just idiotic?
 
User avatar
Gamal
Topic Author
Posts: 2077
Joined: February 26th, 2004, 8:41 am

Re: Quantitative methods in magic

March 13th, 2018, 9:18 am

Not the paper itself but the subject. What we did in the 90-ties and then had some methodological background, we were able to explain in a few words why and what. Their approach is closer to what astrologists did a few hundred years back - some unjustified operations on numbers with nonclear conclusions which could be the opposite. Back to olde good times of magic, ladies and gentlemen.
 
User avatar
katastrofa
Posts: 6123
Joined: August 16th, 2007, 5:36 am
Location: Alpha Centauri

Re: Quantitative methods in magic

March 13th, 2018, 6:27 pm

ISayMoo wrote:
I briefly wondered once if various Deep Learning models which require random inputs (e.g. autoencoders) could learn to predict the next random value generated, thus "cheating". But I doubt it's possible without a humongous training sample and large model capacity.

They could guess the PRNG's seed state based on several numbers. Learning to hack :-) But you do use cryptographically secure PRNGs, don't you.
ABOUT WILMOTT

PW by JB

Wilmott.com has been "Serving the Quantitative Finance Community" since 2001. Continued...


JOBS BOARD

JOBS BOARD

Looking for a quant job, risk, algo trading,...? Browse jobs here...