I occasionally give talks, write papers, and speak on podcasts. Most of the subject matter follows from my professional interests: data science, machine learning, network science, etc.
I’m not an Ethicist — Episode 008 (Guest on The Podcast of Small Differences 2018)
Data Science has become dangerous. You could be hurting other people through the practice of your job, and there’s no reason to try to reason through it all on your own. Guest John Taylor discusses some ethical frameworks as tools to help you with tough issues.
Network Science for Fun and Profit (for Open Source Bridge 2014)
Understanding the relationships between data elements has become increasingly valuable, as LinkedIn, Facebook and Google illustrate. Network science provides a means to understand, explain, predict and otherwise utilize these relationships. I will provide a brief overview of network science, with examples and illustrations using R, focused on providing an entry point to their use for fun and profit.
Machine Learning in the Open (for Open Source Bridge 2012)
Machine learning and data mining methods underlie many exciting products and services, but their underlying workings remain opaque to many, even developers. I will provide a brief tutorial on some of the most important concepts and methods from machine learning and data mining, with motivating examples and illustrations from open source tools. Topics will include data exploration, data preparation, supervised and unsupervised learning methods (including models, patterns, scoring functions, optimization, and search), performance tests, and model evaluation. Particular emphasis will be placed on learning methods and their appropriate use.
Data Science promises to transform ubiquitous and cheap data into insights with the potential for great social, scientific and personal value. However, while many of us have the needed hacking skills and domain knowledge, we might not have a strong background in the agglomeration of formal disciplines that underpin data science methods. I will provide a lightning tour of high level theory, concepts, and tools to extract knowledge and value from data. These are deep and wide subjects, so emphasis will be placed on the high level structures of data analysis problems that point to good solutions.
Computing Spaces (Private Google Project Site) Deprecated
This monograph presents an overview of research on methodology, and argues for an approach unified under category theory. Briefly, it is argued that the computational power of our methods, along with the exact structural features (algebraic, topological, etc.) of the problem and the solution space determine which methods are appropriate in terms of reliability and efficiency. This characterization is, in turn, embedded into category theoretic schemas that aid in classifying learning problems by structures that bridge results across several disciplines (empirical process theory, learning theory, domain theory, and recursion theory).
These papers do not necessarily reflect my current views.
The concern for simplicity is a unifying theme in much of Bertrand Russell’s philosophical works; particularly in his theory of definite descriptions, logical atomism and neutral monism. The latter two theories’ connection with simplicity-considerations will be briefly mentioned, while special attention will be given to the resolution of the three central puzzles in Russell’s ‘On Denoting’. Last, Russell’s justifications for the use of Ockham’s razor will be considered.
Propositional attitudes are, Dennett maintains, here to stay, contrary to the arguments of eliminative materialists. Furthermore, they are not part of “the furniture of the world” as realists maintain, but neither are they mere convenience as instrumentalist see them. Rather, they are real patterns discernable from the intentional stance. What is a stance? What is the intentional stance, in particular? What constitutes a real pattern? Do these notions sufficiently distinguish Dennett’s position from others? How real are real patterns and is their reality problematic?
Steve Austin Versus the Symbol Grounding Problem (co-written with Dr. Scott Burgess) In Proc. Selected Papers from the Computers and Philosophy Conference (CAP2003), Canberra, Australia. Conferences in Research and Practice in Information Technology, 37. Weckert, J. and Al-Saggaf, Y., Eds., ACS. 21-25.
Harnad (1994) identifies the symbol grounding problem as central to his distinction between cognition and computation. To Harnad computation is merely the systematically interpretable manipulation of symbols, while cognition requires that these symbols have intrinsic meaning that is acquired through transducers that mediate between a cogitator and the environment. We present a careful analysis of the role of these transducers through the misadventures of Steve Austin, the Six Million Dollar Man. Putting Steve through a series of scenarios allows us to analyze what role transducers play in cognition.