Artificial intelligence is a machine that mimics a _____ function of human mind

cognitive

Machine learning the study and construction of _____ that can learn from and make predictions on data

algorithms

Dimensionality of a dataset is the _ of the features

sum of the dimensions

Supervised learning is the Data mining task of inferring a function from _ training data.

labeled

Unsupervised machine learning is the machine learning task of inferring a function to describe ____ from "unlabeled" data

hidden structures

Deep learning is a neural network with _____

one or more hidden layers

Artificial neural networks are a _____ on simple neural units

computational model based

An important idea in machine learning

Moderate the updates

Instead of jumping enthusiastically to each new A, we take a fraction of the change ΔA

Keeping some of the precious data

Neurons all transmit an from one end to the other, from the ______________________________

dendrites along the axon to the terminal

These signals are then passed from ___________

one neuron to the other

Signals from are transmitted __________________ along your nervous system to your brain, which itself is mostly made of neurons too

specialize sensory neurons

Artificial Neuron

Receives input from sources

Computes the weighted sums

Passes through an activation function

Sends the signal to m succeeding neurons

Neurons don’t react

readily

neurons suppress the input until it has grown so large that it _______

triggers an output

Threshold that must be reached before any other _________

outputs are produced

The electrical signals are collected by the ______ and these combine to form a stronger ______

dendrites, signal

If the signal is strong enough to pass the ____

threshold

The neuron ___ a signal down the ____ towards the terminals to pass onto the next neuron’s dendrites.

fires, axon

Each neuron takes input from many before it

Provides neruon to many more

Propagate signals __ from the input to the output layers

forward

Propagate the error _ from the output back into the network

backwards

What’s the link between this really cool gradient descent method and neural networks?

If the complex difficult function is the error of the network

Going downhill to find the minimum means we’re minimizing the error

We’re improving the network’s output. That’s what we want!