Approaches to A. I. Thinking like humans Cognitive science Neuron level Neuroanatomical level Mind level Thinking rationally Aristotle, syllogisms Logic.
Post on 13-Dec-2015
219 Views
Preview:
Transcript
Approaches to A. I.Thinking like humans• Cognitive science• Neuron level• Neuroanatomical level• Mind level
Thinking rationally• Aristotle, syllogisms• Logic• “Laws of thought”
Acting like humans• Understand language• Play games• Control the body• The Turing Test
Acting rationally• Business approach• Results oriented
Human Rational
Thinking
Acting
(Artificial) Neural Networks
• Biological inspiration• Synthetic networks• non-Von Neumann• Machine learning• Perceptrons – MATH• Perceptron learning• Varieties of Artificial Neural Networks
Brain – Network of Neurons
Each neuron has on average 7,000 synaptic connections with other neurons.A neuron “fires” to communicate with neighbors.
Animal Neural Architecture
von Neumann• Separate processor and
memory• Sequential instructions
Birds and bees (and us)• Each neuron has state and
processing• Massively parallel,
massively interconnected.
The Percepton
• A simple computational model of a single neuron.
• Frank Rosenblatt, 1957
• The entries in are usually real-valued (not limited to 0 and 1)
How to “program” a Perceptron?
• Programming a Perceptron means determining the values in .
• That’s worse than C or Fortran!• Back to induction: Ideally, we can find from a
set of classified inputs.
Perceptron Learning RuleInput Output
x1 x2 x31 if avg(x1, x2)>x3, 0 otherwise
12 9 6 1-2 8 15 03 0 3 09 -0.5 4 1
Training data:
Valid weights: 𝑤1=0.5 ,𝑤2=0.5 ,𝑤3=−1.0 ,𝑏=0
Perceptron function: { 1 if 0.5 𝑥1+0.5 𝑥2−𝑥 3−0>00o therwise
Varieties of Artificial Neural Networks
• Neurons that are not Perceptrons.• Multiple neurons, often organized in layers.
Neural Networks
• Alluring because of their biological inspiration– degrade gracefully– handle noisy inputs well– good for classification– model human learning (to some extent)– don’t need to be programmed
• Limited – hard to understand, impossible to debug– not appropriate for symbolic information processing
top related