| Predictions | Model | Archive | Authorship |

Neural Network Model Information

(This is an archive of the first version of the model utilized in 2018.)


Source: United States Geological Survey (USGS)
Training Set: 1973-01-01 to 2016-12-31
Test Set: 2017-01-01 to 2017-12-31
Update Set: 2018-01-01 to 2018-02-28
Seed Set: Last 1 Day (Previous 24 Hours)
Temporal Resolution: 1 Day (24 Hours)
Spatial Resolution: 2 Degrees Latitude and 2 Degrees Longitude
Details: Earthquakes were reorganized into tensors, with each slice representing a single day and consisting of a matrix with the rows and columns representing approximate latitudes and longitudes. Each position in the matrix was filled with a magnitude determined by converting all the earthquakes occurring at that approximate location and time from logarithmic scale magnitudes into linear energies, summing them together, and then converting the resulting value back into the logarithmic scale.


Long Short Term Memory (LSTM) Recurrent Neural Network (RNN)
Hidden Layers: 5
Neurons Per Hidden Layer: 512 nodes
Timesteps: 1
Epochs of Training: 1
Loss Function: Asymmetric (Exponential Or Logarithmic)
Library: Keras with Theano backend
Notes: Model is fully Stateful and utilizes an Online Learning schedule. Utilizes Layer Normalization and Residual Skip Connections as well as Golden Ratio scaling of the activation functions, and Golden Ratio Conjugate scaling of the gradient norms. No dropout or stochastic timeskip.


Prediction Magnitudes:

Either Prediction or Actual Magnitude Is 0.0 Or Greater

Either Prediction or Actual Magnitude Is 5.0 Or Greater

(Note: Some statistics cannot be properly assigned because of the lack of true negatives. Also, these numbers are somewhat biased due to the sparsity of predictions above 5.0)