Thanks Lee. Appreciate the pointer to Jay Alammar - great visualizations. I liked and subscribed to his YouTube channel. His video on word embeddings is also quite well visualized. The word guessing game of "Twenty Questions" is also a way of thinking about words as a vector of numbers corresponding to "yes" or "no" or finer shades to specific questions. An early AI that was quite impressive (built into toys) used this type of approach is described here: https://en.wikipedia.org/wiki/20Q
Thanks Lee. Appreciate the pointer to Jay Alammar - great visualizations. I liked and subscribed to his YouTube channel. His video on word embeddings is also quite well visualized. The word guessing game of "Twenty Questions" is also a way of thinking about words as a vector of numbers corresponding to "yes" or "no" or finer shades to specific questions. An early AI that was quite impressive (built into toys) used this type of approach is described here: https://en.wikipedia.org/wiki/20Q
Jim, thanks for the pointer to the 20Q stuff -- I wasn’t aware of that.