Center for Mind, Brain, and Culture

Neuroscience Workshop/Lecture (4 of 5) | Phil Wolff | The Large-Scale Structure of the Mental Dictionary: A Data Mining Approach Using Word2Vec, t-SNE, and GMeans

Episode Summary

Workshop/Lecture | Phil Wolff | The Large-Scale Structure of the Mental Dictionary: A Data Mining Approach Using Word2Vec, t-SNE, and GMeans

Episode Notes

Advancements in machine learning and data mining have already led to amazing breakthroughs in the natural sciences, including the unlocking of the human genome and the detection of subatomic particles. Such techniques promise to wield a similar impact on the study of mind. In my talk I will discuss how the large-scale structure of the human mental lexicon, roughly 50,000 words, can be recovered from billions of words at a level of resolution that includes the differentiation of word senses. Central to this effort are several machine learning and dimensionality reduction techniques, including deep learning, t-Distributed Stochastic Neighbor Embedding (t-SNE), and the clustering technique called GMeans. In addition to the extraction of the mental lexicon, I will discuss how an approach to topic modeling, based on neural networks, might be used to partially automate the process of theory generation. I also raise implications for research on physical and mental wellbeing. NEUROSCIENCE WORKSHOP: Dimensionality Reduction Friday, October 30, 2015 Saturday, October 31, 2015