Myopic histories and AI's culture of hyperbole

Histories

• Schmidhuber, Jürgen. "Deep learning in neural networks: An overview." Neural networks 61 (2015): 85-117.
• Bishop, J. Mark. "History and philosophy of neural networks." in Computational Intelligence (2015): 22-96.
• McCorduck, Pamela. "Meat Machines" in Machines who think: A personal inquiry into the history and prospects of artificial intelligence. AK Peters/CRC Press, (2009).
• Olazaran, Mikel. "A sociological study of the official history of the perceptrons controversy." Social Studies of Science 26, no. 3 (1996): 611-659.


• Many more unofficial histories precede textbooks, review papers and blog posts

Schmidhuber's overview

Begins section on early neural networks with McCulloch and Pitts in the 1940s but gives passing reference to history of methods;

"In a sense [neural networks] have been around even longer, since early supervised NNs were essentially variants of linear regression methods going back at least to the early 1800s (e.g., Legendre, 1805; Gauss, 1809, 1821)"

Allows Schmidhuber to admit the origin of methods but maintain the simulation of the brain as the central narrative

The McCulloch-Pitts neuron

The McCulloch-Pitts neuron followed from a number of axioms derived from the existing theories of neurophysiology and empirical evidence.

• The activity of the neuron is an 'all-or-none' process.
• A certain fixed number of synapses must be excited withing the period of alatent addition in order to excite a neuron at any time, and this number is independent of previous activity and position of the neuron.
• The only significant delay within the nervous system is synaptic delay.
• The activity of any inhibitory synapse absolutely prevents excitation of the neuron at that time.
• The structure of the net does not change with time.