AI-powered
podcast player
Listen to all your favourite podcasts with AI-powered features
Is There a Negative Association With Building Systems by Hand?
In the traditional deep learning approach, like the knowledge of the system is kind of stored in weights and ambedding and things like that. It sounds that your representations are more external. Does your work have this kind of traditional view of a model, and how does that relate when you're building systems around your representations? Like, what do they look like? Are they similar to what we might be used toit with deep learning and machine learning? Or are they very different in some ways? Were many ways they would look like structures of kinds of models and types of systems and tures that were way more popular,. perhaps, in the 19 seventies and 19 eighties.