ITM 740 Terminolgy and Theory
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!