Reference Information:
Title: Addressing the Problems of Data-Centric Physiology-Affect Relations Modeling
Authors: Roberto Legaspi, Ken-ichi Fukui, Koichi Moriyama, Satoshi Kurihara, Masayuki Numao, Merlin Suarez.
Venue: IUI’10, February 7–10, 2010, Hong Kong, China.
Summary: Using a data-centric approach, the researchers of this paper attempted to come up with a way to use machine learning in order to define so called "affective states". Using an EEG helmet, the researchers attempted to classify, and define the different emotions that the wearer was experiencing.
One of the main problems that they ran into was the size of the data set. Existing emotion sensing algorithms are general of O(n^2) or O(n^3) complexity, and so using a large data set slows the process down. This is a problem when attempting to sense something as fluid and changing as emotion, which can move on to something new before your previous analysis is completed.
Throughout the paper, they suggest several different changes to the algorithms used, and change in data set size, which may help improve the capabilities of the machine.
Discussion: This was the hardest paper I've read in the class. It was not designed for someone with an inexperience in the subject, and I could hardly understand what they were talking about. I'm sure their research is interesting, but I'm not sure if anyone is going ever know what they actually did, judging from this paper.
I agree that this paper was incredibly difficult to read. I would assume that this would be because they have a lot of special jargon for scientifically classifying emotions, but they should have worked a lot harder on making it somewhat readable.
ReplyDeleteI can certainly see the problem with the size of the data sets. I haven't read this paper, but I'm curious as to what method of machine learning the authors are using (neural networks, etc.) and what their ultimate goal with the research is.
ReplyDeleteThis paper was ridiculously confusing. I'm surprised it got published given how easy other articles we've read are to understand.
ReplyDelete