Himanshu Joshi

PhD Candidate in Artificial Intelligence and Autonomous Agents, University of Southern California Department of Computer Science
Research Assistant, Institute of Creative Technologies

Hi! This is a text-positive environment, that I have shamelessly plagiarized in the style of Johnathan May's Web space.

I am working on fusing speech with natural language in the Sigma 'cognitive architecture.' The main aim of my thesis is to develop a theory of spoken language parsing, along the lines of a cognitively plausible spoken language model.

In my former life, I wrote portions of operating systems, device drivers and network protocol layers for cellular phones. Most notably, I wrote the LTE connected mode searcher, responsible for maintaing calls in the face of fluctuating radio conditions. So, if your iPhone or Android drops a call while you are driving, blame the people maintaining my code.

I should update my resume more often.

This is my advisor's page.

Someday, there will be a Joshi prior. Until then, we have the Demski prior.




"Continuous Phone Recognition in the Sigma Cognitive Architecture." (Joshi, H., Rosenbloom, P. S., & Ustun, V.), To appear in Biologically Inspired Cognitive Architectures (BICA), 1, 2016. pdf video


"Isolated word recognition in the Sigma cognitive architecture." (Joshi, H., Rosenbloom, P. S., & Ustun, V.), Biologically Inspired Cognitive Architectures (BICA), 10, 1-9, 2014. pdf video



Improved parsing F1, precision and recall of CKY parsing using word embeddings. report