Big Data - Exam II

  1. neural computing
    a pattern-recognition methodology for machine learning with the ability to "learn" from data, and the ability to generalize
  2. artificial neural network (ANN) or neural network
    the resulting model from neural computing used for pattern recognition, forecasting, prediction, and classification
  3. weights
    provide the relative importance of each input in an artificial neural network
  4. Artificial Neural Network
  5. neurons
    the main processing elements of a neural network
  6. node
    ANN equivalent of a soma
  7. input
    ANN equivalent of a dendrite
  8. output
    ANN equivalent of an axon
  9. weight
    ANN equibalent of a synapse
  10. soma/node
    middle of a neuron
  11. dendrites/input
    the portion of a network that receives signals
  12. axon/output
    the portion of a network responsible for sending signals to other cells
  13. synapse/weight
    the portion of a network that is able to alter signals by increasing or decreasing the strength of the connection
  14. transformation (transfer) function
    the function in an ANN that is responsible for the signal to "act" or "not"
  15. feedforward-backpropagation paradigm (packpropagation)
    allows all neurons to link the output in one layer to the input of the next layer, but does not allow any feedback linkage - the most commonly used network paradigm
  16. network
    organized neurons
  17. topologies
    network patterns of neurons
  18. artificial neurons
    the processing elements (PE) of an ANN which receives inputs, processes them, and delivers a single output
  19. hidden layer
    a layer of neurons that takes input from the previous layer and converts those inputs into outputs for further processing
  20. processing elements
    the "neuron" - performs the summation function
  21. parallel processing
    many processing elements perform their computations at the same time
  22. input
    represents a single element (ex: sales price) to be considered
  23. inputs
    several types of data, such as text, pictures, and voice, can be used as ____.
  24. output
    the solution to the problem - the purpose of the network computations - often in a 0 (no) or 1 (yes) format
  25. connection weights
    the key elements of an ANN - expfress the relative strength / mathematical value / importance of the input data or the many connections that transfer data from layer to layer
  26. weights
    through repeated adjustments of ____ a network learns
  27. weights
    store learned patterns of information
  28. summation function
    computes the weighted sums of all the input elements entering each processing element - multiplies each input value by its weight and totals the values for a weighted sum
  29. transformation (transfer) function
    combines (i.e., adds up) the inputs coming into a neuron from other neurons/sources and then produces an output
  30. sigmoid (logical activativation) function
    non-linear S-shaped transfer function in the range of 0 to 1
  31. threshold value
    a hurdle value for the output of a neuron to trigger the next level of neurons
  32. supervised learning
    neural network learning where a sample training set is used to "teach" the network about its problem domain
  33. learning algorithm
    determines how the neural interconnection weights are corrected due to differences in the actual and desired output for a member of the training set
  34. supervised learning
    type of ANN learning used for prediction-type problems - an exemplary data set is used against a learning algorithm that adjusts as it examines its own outcomes against predetermined preferred outcomes - data is labeled (good or bad)
  35. unsupervised learning
    type of ANN learning used for clustering-type problems - learns patterns by repeatedly examining data that is not labeled
  36. learning rate
    a parameter that controls the rate of learning (too high or too low a rate causes issues)
  37. unserpervised learning
    network does not try to learn a target answer, instead, it learns a pattern through repeated exposures
  38. 1. collect organize & format the data
    2. separate ate into training validation & testing sets
    3. decide on a network architecture
    4. select a learning algorithm
    5. set network parameters & initialize their values
    6. initialize weights & start training (& validation)
    7. stop training, freeze the network weights
    8. test the trained network
    9. deploy the network for use on unknown new cases
    steps of the deployment process of an ANN
  39. sensitivity analysis
    the inputs are perturbed (changed) within allowable value ranges while the relative change on the output is measured/recorded
  40. sensitivity analysis
    developed to help users understand how the model performs - conducted on a trained ANN - results illustrate the relative importance of input variables
  41. logistical regression
    the regression type that can handle a categorical outcome variable
Card Set
Big Data - Exam II
Artificial Neural Networks