One of the guiding principles here at the AHA!Institute is that each person, regardless of their background or profession can make a meaningful contribution to the collective effort. Physicists have something to learn from Janitors and Janitors have something to learn from Physicists. My personal understanding of many of the weightier ideas expressed here may not always be immediate, but by staying present to the conversation quite often the underlining principles reveal themselves. Lately, I’ve been returning to the conversation many of the programmers are having in their attempt to construct the Autognome, the "learning" and “thinking” component of Ai3’s flagship Online Knowledge Management and Community Building Tool. The conversation as I interpret it goes something like this: “We are training the Autognome to look at sets of raw data/information. The Autognome logically forms "signs" with respect to this data/information and then groups those signs in terms of similarity into "classes (or categories)". These “signs” and "classes" form character sets for a "tentative language". The Autognome uses this language to interpret new data and then check its results for accuracy with respect to experience." With all of this in mind I have set out to apply a similar line of thought to my approach toward tying flies for competitive fly fishing. I will look at a set of data (fish feeding behavior); Form signs from that data (tie flies to invoke positive reactions); Cluster data (group positive reactions); Then apply those signs to new data, record, and modify for accuracy (use flies in different settings). This may not be Noble Prize winning work, but it could mean quite a bit to the angler. The basis for my quantitative work will stand on the theory of “Suggestive” rather than “Imitative” fly tying. There are generally 2 philosophical houses of fly tying: the first sets out to key reaction strikes based on stimuli such as shape, color, and size while the second seeks to reproduce fish forage as closely as humanly possible with artificial materials. Please look for updated reports on this “Query” in the upcoming weeks and months.
No comments:
Post a Comment