|
Greetings to all
Could someone please recommend book/s and/or readings for Artificial Intelligence and Signal Processing.
Thank you very much.
A |
|
|
|
 |
|
| For signal processing take a look at the books by Steven Kay (statistical signal processing). |
|
|
 |
|
HankScorpio,
Thank you very much for taking the trouble.
I'll definitely have a look at the book and once again sincere thanks.
A |
|
|
|
|
| |
|
MrMagoo
|
|
| Total Posts: 188 |
| Joined: Jan 2008 |
| |
|
Follow my links below :
If you want to learn from scratch and put serious study time, you can understand the basic framework, then slowly jump to the state of art.
hardly u can get a better beginner's online stuff.
I took these 2 for fun last year. I liked, even though i knew the subject at a higher level than presented.
beginner's AI class (co-taught by P.Novig, co-author of "the AI" book.)
Machine Learning basics
I barely have time now, but i would seriously consider any coursera courses :
this one may interest you.
digital signal processing |
"One who says it can't be done should not interrupt the person doing it."
|
|
|
|
|
frstwrldprblm- Thank you very much for the recommendation.
I have got the book, just now. From the look of first few chapters, it does really look right kind of book for me.
Thank you once again for your very kind help.
A |
|
|
|
|
MrMagoo- Thank you very much for the suggestion.
Before posting the message here, I, indeed, was looking at Coursera courses. I am having a look at the course available over there once again.
I also have found AI online course notes from P.Novig. I will, for sure, be studying the notes and watching the videos which are available on the portal. The notes and video seems me to be very interesting and contents are what I was looking for.
Thank you once again for taking time in replying the message.
A |
|
|
|
|
bluelou
|
|
| Total Posts: 63 |
| Joined: Jan 2009 |
| |
|
I can't speak to AI more broadly but I just finished the Coursera ML course and I plan to take the Probabilistic Graphical Models (PGM) course in Sep. The ML course is taught very well. It's very strong on the intuition behind various ML techniques but it's a bit light on the math/technical aspects underlying the subject. I hear the PGM course is more in-depth and very time consuming.
For an ML book where you can follow along from the code figures (plus Python code at website) you may want to look at Machine Learning by Stephen Marsland. For, AI, I would guess that you want to cover particle filters. This book mentions them but only covers Kalman filters in detail. |
Je suis ce que je suis, et c'est tout ce que je suis -Popeye
|
|
|
gax
|
|
| Total Posts: 3 |
| Joined: Apr 2011 |
| |
|
| Take a look at: Pattern Theory: The Stochastic Analysis of Real-World Signals by David Mumford and Agnès Desolneux. It's a beautiful book. |
|
|
|
|
jslade
|
|
| Total Posts: 745 |
| Joined: Feb 2007 |
| |
|
One of the problems with all the AI or machine learning books I've looked at (including my 2nd edition of Norvig, which I would not recommend to anyone but my enemies): they're more interested in classification problems than time series problems. Most ML books have a pretty cursory treatment of TS, generally when addressing Hidden Markov or reinforcement learning type things.
There is also the matter of noise in general, let alone time dependent noise, which seems an afterthought to most AI/ML guys, unless they're covering Kalmany things.
I guess if you know what you're doing and are well grounded in signal processing techniques, it doesn't matter so much, but it can be confusing, and you can get into all kinds of trouble.
Maybe that Mumford book is better at covering such issues; it certainly looks really good. |
"Learning, n. The kind of ignorance distinguishing the studious." |
|
|
quantie
|
|
| Total Posts: 861 |
| Joined: Jun 2004 |
| |
|
Duda -Hart
& Fukinaga are the classic references.
But as Jslade mentions the problem with financial data is it is very noisy so a lot of the techniques are an overkill. From talking to people who are successfully doing this it appears to me that linear models are all that you need , you need to spend time preprocessing/cleaning your data but all this fancy machinery can kill you. (Caveat emptor we don't do anything using ML right now so a practitioner here will quickly correct me.) |
|
|
|
|
|
| bluelou- Thank you very much for taking time in offering the suggestion. |
|
|
|
|
| bluelou- Thank you very much for taking time in offering the suggestion. |
|
|
|
|
|
| gax- Thank you very much for your suggestion. |
|
|
|
|
| jslade- Thank you very much for your views and thank you very much for taking time in replying the post. |
|
|
|
|
jslade
|
|
| Total Posts: 745 |
| Joined: Feb 2007 |
| |
|
FWIIW, I bought the Kay books HankScorpio suggested: these are wonderful. If every book were like these, I'd be out of a job. They're actual cookbooks. Thanks HS!
The book suggested by gax (Pattern Theory by Mumford) is as described; beautiful. I don't think it is very practical, but one could learn a lot from it. It's kind of a philosophy book. |
"Learning, n. The kind of ignorance distinguishing the studious." |
|
|
mtsm
|
|
| Total Posts: 94 |
| Joined: Dec 2010 |
| |
|
| These are great books, agreed, but they are not signal processing books in the classical sense and do not really represent what most people understand by the notion of signal processing. Feel free to contradict, as I am by no means an expert... The scope of these is remarkably similar to many econometrics texts. |
|
|
|
|
|
jslade, you're welcome.
mtsm, would you care to elaborate further on what you believe people understand when they refer to SP in the "classical sense"? Perhaps you could provide some books for examples.
Maybe it's because I'm an EE by training, but what is described in Kay's books would be what I'd be thinking of when referring to SP. |
|
|
|
|
I concur with jslade about the books by Steven Kay.
The chapters are quiet straightforward, the language is very easy to understand and most importantly the problems at the end of chapter are quiet testing which is what I was looking for.
Also, a large number of figures have been used to explain important concepts.
To sum up, the books are indeed excellent.
Thank you once again, HankScorpio, for recommending the books.
A |
|
|
|
|
mtsm
|
|
| Total Posts: 94 |
| Joined: Dec 2010 |
| |
|
| Well I was more thinking along the lines of empirical signal analysis, filtering, spectral analysis, time-frequency analysis, wavelets, you know the usual stuff. There are a lot of DSP books out there. As a failed physicist I am thinking for example of the Bendat, Piersol book, 'Analysis of Random Data', that is not to cite any digital signal processing references that I don't really know about. |
|
|
|
|
one of the best practical books on DSP (sans wavelets)
.. and it's free.
http://www.dspguide.com/
For practical wavelets, I recommend--
http://www.amazon.com/Wavelets-Scientific-Applications-Advanced-Mathematics/dp/1584887451/ref=sr_1_2?ie=UTF8&qid=1344236083&sr=8-2&keywords=wavelets
This is quite good for practical financial time series SP--
http://www.amazon.com/Neural-Hybrid-Algorithms-Series-Prediction/dp/0471130419/ref=la_B001HMTV6Y_1_2?ie=UTF8&qid=1344236599&sr=1-2
(don't be fooled by the NN title-- it covers practical filtering/econometrics concepts as well)
---------------------------------------
The above are all light on math, heavy on practical and intuitive use.
They also include programs (or practical code) to implement the concepts, rather than just descriptive algorithms.
|
|
|
|
|
|
etrader12
Thanks for the recommendations.
Managed to find all the books except the last one. Hopefully, I can find it in another library.
Thank you once again.
A |
|
|
|
