Subscribe to DSC Newsletter

Why do computer science books talk bout decision trees but not about logistic regression nor naive Bayes?

I was reading the 2004 edition of "Computer Science Handbook", published by Chapman & All/CRC, in collaboration with the ACM association. It has 2,752 pages, weights 5.8 pounds according to Amazon.com, and has a 56-page index with about 6,000 keywords. Yet naive Bayes and logistic regression are not listed, just to mention 2 popular data mining keywords.

It is incredibly surprising, in the 21-st century, to see that disciplines such as data mining, computer sciences, and statistics appear to be almost totally separated by a wall which in several ways reminds me of the the Berlin Wall.

Integration of techniques from various fields into a common knowledge  will be one of the major drivers of scientific progress in the next 20 years.

Views: 275

Comment

You need to be a member of AnalyticBridge to add comments!

Join AnalyticBridge

Comment by David Dodds on February 9, 2011 at 1:28pm

PART 2

I have heard very little from anyone else since about uptake of those ideas. An "integration..." project I have been looking at the last couple of years is combining SVG (via its metadata element), MathML, and OWL ontologies (especially ones about mathematics); such that the SVG "picture" can 'understand' the mathematics involved in it. Perhaps a second derivative was involved in calculating the animation of part of the SVG picture. The second derivative in the SVG file is represented by executable computer code which performs the mathematical process. The SVG metadata element for that SVG file could have an id reference to the part of the SVG picture that was animated by the math code and it could also contain an RDF pointer / reference to the entry for 'second derivative' in an (external) OWL mathematics ontology. A Topic Map can conceptually link all these things together. What results is 1) an SVG animation, 2) which uses an executable form of mathematics knowledge [computer programming to perform second derivative calculation], 3) a conceptually linked reference (RDF link and Topic Map) between the animated part and an ontological depiction (in OWL) of the concept 'second derivative'. By applying a reasoner software (Description Logic) to this the computer can 'realize / see' that the motion in the picture is a mathematically based thing, specifically the concept of 'second derivative'. This contrasts with the computer merely dumbly executing code and having no clue about the motion or the mathematical concepts which are part of that motion. Richard's comment about 'integration...' is crucial for development towards real advancement not only in computing but in many other technical endeavours. There still today remain many bastions, silos, protected turfs which prevent rapid progress.

Comment by David Dodds on February 9, 2011 at 1:24pm
Richard said "Integration of techniques from various fields into a common knowledge  will be one of the major drivers of scientific progress in the next 20 years."
In order for Kurzweil's Singularity to happen ever, let alone at the date he gave, exactly what Richard said must actually occur. The Singularity is an adult human level of intelligence in a machine. Look at the capabilities of Alex the Grey Parrot (Pepperberg) [there are useful videos of Alex on utube]. Alex's brain is the size of a walnut. Even with 10nm scale nanocircuits computers still lack the programming to make that speed even as  clever as bumble bee. It is because the *intrinsic* capabilities of that nanocircuitry computer pale into insignificance compared to Alex's walnut sized processor. The difference is his (processor / brain) intrinsic capabilities are vastly greater in capability. As Richard said "Integration of techniques from various fields into a common knowledge ..." is the only way we will approach Alex's competency let alone serious move towards achieving anything like adult human level of intelligence in a machine. An example of what "integration..." can accomplish is IBM's WATSON computer system. Various fields including linguistics and computing science were integrated into one system, yielding an advancement in field-able machine intelligence. WATSON appears not to be conscious nor (self) 'creative', apparently not able to reflect on either its functioning, how it goes about doing its processing, nor of generating new ways of going about its business. (I'm not faulting IBM.) John C. Lilly wrote about programming and meta-programming. He had some very valuable insights about how to go about achieving second order metaprogramming. In the time since his publication I have seen no system which does, or even project which attempts to do, what Lilly discussed. I met with Lilly at his lab in California and discussed metaprogramming. I have heard very little from anyone else since a

On Data Science Central

© 2020   AnalyticBridge.com is a subsidiary and dedicated channel of Data Science Central LLC   Powered by

Badges  |  Report an Issue  |  Privacy Policy  |  Terms of Service