training_nnet: Neural network training

In PabloMBooster/vmsR: For analysing VMS data

Neural network training

Description

Training a artificial neural network to identify fishing sets in vms data. The following variables are required for neural network training: hour, change of acceleration and change of course.

The neural network is trained on the basis of the next job (joo 2011). For the validation of the model check the percentage of successes and for increase the precision of the model it is necessary to increase the number of loops.

We use the nnet package and any argument can be modified to improve the performance.

Usage

training_nnet(data, directory, formula, neurons = 4, loops = 50, thres_min = 0.4, # dossier.0, directorio,
                           thres_max = 0.6, MSE_max = 0.04, prop_train = 0.75, T1 = 180, T2 = 360,
                           linout = FALSE, entropy = FALSE, softmax = TRUE,
                           censored = FALSE, skip = FALSE, rang = 0.7, decay = 0,
                           maxit = 100, Hess = FALSE, trace = FALSE, MaxNWts = 1000,
                           abstol = 1.0e-4, reltol = 1.0e-8)

Arguments

data	data sets
directory	the working directory
neurons	numbers of neurons
nb_loop	numbers of loops
thres_min	minimun threshold
thres_max	maximun threshold
MSE_max
prop_train
T1
T2

References

- Joo R., Bertrand Sophie, Chaigneau Alexis, Niquen M. (2011). Optimization of an artificial neural network for identifying fishing set positions from VMS data : an example from the Peruvian anchovy purse seine fishery. Ecological Modelling, 222 (4), 1048-1059. ISSN 0304-3800. http://www.documentation.ird.fr/hor/fdi:010053066.

Examples

head(data_vms)

loops = 2
neurons = 4
MSE_max = 0.04

# calibration
nnet_out <- training_nnet(data = data_vms, directory = getwd(), neurons = neurons, MSE_max = MSE_max, loops = loops)

# prediction
data_vms$Calas <- predict_sets(data = data_vms, directory = getwd(), loops = loops)

# map
require(maps)
plot(data_vms$Lon, y = data_vms$Lat, cex = 0.4, pch = 16, xlab = "lon", ylab = "lat")
points(data_vms$Lon[data_vms$Calas == 1], y = data_vms$Lat[data_vms$Calas == 1], col = 2, cex = 0.4)
map("worldHires",fill=T, myborder = FALSE, add = TRUE, col = "khaki1")
box()

PabloMBooster/vmsR documentation built on June 29, 2023, 11:16 a.m.