Loading...
Please wait, while we are loading the content...
Similar Documents
Extending Markov Chain Monte Carlo with People to categories defined on sets of discrete items
| Content Provider | Semantic Scholar |
|---|---|
| Copyright Year | 2011 |
| Abstract | A recently-developed method known as Markov Chain Monte Carlo for People (MCMCP) has been shown to be successful at measuring how people represent categories. The original MCMCP method could only be used for categories defined over stimuli that can be described by a fixed set of parameters. However, the objects that belong to real world categories – such as images or documents – are often not easy to describe in parametric terms. Here we introduce an extension we call discrete-MCMCP, which can be used to estimate people’s category representations over an arbitrary discrete set of stimuli. This significantly extends the scope of the method, making it possible to apply it to large image databases, sets of documents, and other stimuli that are hard to parameterize. We present experimental results showing that performance is similar to that of the original method when applied to a set of stimuli for which a parametric representation can be found, and demonstrating the flexibility of discrete-MCMCP by estimating natural categories using a set of 4000 images drawn from online databases. |
| File Format | PDF HTM / HTML |
| Alternate Webpage(s) | http://cocosci.berkeley.edu/annehsu/papers/nips2011_MCMCP_finalfinal.pdf |
| Language | English |
| Access Restriction | Open |
| Content Type | Text |
| Resource Type | Article |
