## Elements of Interactive Cognition (Perception):

- Environment
- Sequential Parsing
- Prediction
- Interaction with other systems
- Recurrent (connections)
- Memory
- Parallel/Distributed Processing: not clear whether this is for computational advantages or for some other reasons

**Formal Models:**

- Environment:
- Labeled examples: a sequence of (X
_{1}, l(x_{1})), … (X_{n}, l(x_{n})) - Two models:

(a)**PAC (probably approximately correct)**:

Assume distribution D (unknown) on input examples . After seeing m labeled examples (x, l(x)) where x ~ D, a -PAC algorithm finds a hypothesis h with prob. such that:

(on a new input y ~ D, we should be able to guess l(y))

(b)**Mistake-bound**:

Given an example x, you predict l(x) right away, and are then told the correct answer. The complexity of an algorithm in the mistake-bound model is m, if for any sequence of examples (of any length), the total number of mistakes made by the algorithm is at most m.

- Labeled examples: a sequence of (X
- Prediction:
- Max Likelihood Estimators: find a model that most likely generated the data
- Bayesian Estimation: use priors to make decisions

- Recurrent (connections):
- Recurrent Neural Network (RNN): contains connections that are not feed-forward:

- Recurrent Neural Network (RNN): contains connections that are not feed-forward:

## Project Example:

**Invariant**: Parsing a letter

- Environment and Sequential Parsing: I see a letter as a sequence of strokes

- Prediction: As someone starts writing a stroke, you can predict which letter it is

- Interaction with other systems:
- Motor System
- Sound

- Recurrent: Use feedback from “learned” letters to predict

- Parallel/Distributed Processing:

Advertisements