Autonomous Learning Systems: From Data Streams to Knowledge in Real-time

Description

Autonomous Learning Systems: From Data Streams to Knowledge in Real-time - Kindle edition by Angelov, Plamen. Download it once and read it on your Kindle device, PC, phones or tablets. Use features like bookmarks, note taking and highlighting while reading Autonomous Learning Systems: From Data Streams to Knowledge in Real-time. Review: Innovative approach appearing to hold significant promise - I am a PhD student in Engineering and am looking for non-parametric, autonomous, evolving algorithms for ontology learning and semantic information extraction. A key to this is autonomous learning and throughput (I do not attend the university where the author is employed so this is not a situation where I'm reviewing a textbook for a class instructor). The approaches outlined in this book appear to be very interesting primarily in the innovative and fresh approach to a number of computational intelligence problems that enable autonomous learning and without the need for a priori models or parameters. I am pursuing the application of these approaches further for both my research plus others; of course no results yet but the autonomous learning aspect appears very interesting. Software is available via the startup company for a nominal fee (haven't tried the software yet). I believe the book is clear and communicates the subject in a manner that the chapters build upon each other. While Dr. Angelov does a good job of laying out the basics in statistics, clustering, and fuzzy logic, in general the reader will benefit from a background in these areas so that the material is more approachable and readable. If the reader does not have such a background the basics are in the book but these background subjects are so broad that it is impossible to fit all the necessary details into just one book. Since this is a first edition, like any first edition, it may need a bit of clean up for the next edition. I'd like to add my input on a couple of EXAMPLE items that may be of value for future editions - none of these appear to me to be so significant as to materially detract from the book's value: 1. Notation explaination: in some situations the notation for equations is not complete or refers to a variable found much earlier in the text. It would be more convenient for the reader (example - equation 5.8 - what is sub-notation 'm'?) 2. Example data and density calculations with multiple clusters: I got a bit lost on the ideas behind the density calculations for clusters - does a point belong to more than one cluster and how does this impact density calculations for each cluster? A simple but real-world set of data and example calculations would be very helpful to validate how the local and global density is calculated. The pseudo-code in appendices should help but havent' gotten to that yet. This could be accomplished in the latter chapters devoted to real-world applications. 3. Criteria for degree of cloud membership: unclear to me in equation 4.14 and Figures 2.5 and 2.6 what specific, concrete criteria to use for determining cloud membership, which appears to be very critical when applying the criteria specified in section 5.2.5 for autonomous data partitioning. Pseudocode in appendices may address this. But, the level of innovation appears to be so high that I gave it the highest score possible. To me that is the core value. The main benefit of reading this book to me is to look at a problem through the lense of a different approach that may enable a much higher level of autonomous learning, and if it looks like it will solve the problem then dig into the nitty gritty of applying the approach. Review: It helps in my research