Lecture2 vision

Recognition by components model:

To view that an object is represented as an arrangement of simple 3-D shapes called geons

prototype Model

object perception involves a comparison of the stimulus with ideal, abstract example

Alternative models of perception

mindfulness is largely about seeing the" suchness" of things, that is, seeing things directly without conceptual filters ( assumptions)

Template matching model:

object perception involves a comparison of the stimulus with set of templates or specific patterns stored in memory

Feature analysis Model

discrimination of objects is based on small number of characteristics of stimuli

Top-down processing

information processing guided by higher-level processes, such as our beliefs, expectations, and memories

Bottom-up processingg:

analysis of information coming from stimuli through sensory receptors

Self-fulfilling propheciespeople generally think that it is our experiences and perceptions that create our beliefs but often, it is actually our beliefs that create our experiences and perceptions.

The Pygmalion effect

study found that students who were ( randomly) labeled intellectuals showed significantly greater gains in IQ and academic performance

Visual agnosia:

inability to recognize/identify visual objects despite relatively food visual perception. Ex: v

Visual neglect syndrome or unilateral spatial negelct:

Tendency to ignore- or to be unaware of - information on one half of visual field, usually the left side.

Capgras syndrome:

Characterized by belief that family and/or friends are imposters

Functional blindness:

Unexplained vision loss with no organic basis, ex: Cambodian women who had witnessed horrible war atrocities became either partially or wholly blind.

Conversion disorder:

Lose use of a limb or experience

First (input)layer of network:

Starts with bunch of neurons or nodes corresponding to an array of 28×28 pixels in the image

Second layer(the first “Hidden Layer”)

each neuron in the 2nd layer might pick up on whether there is an edge in one particular region

Third layer( second hidden layer)

when we recognize digits, we piece together various components

These subcomponents are made up of the various edges from the second layer

How many hidden layers are needed?

One hidden layer is sufficient most of the time

Why are more hidden layers not necessarily better?

Cause the network to overfit the training set.

backpropagation