Skip to content
Snippets Groups Projects
Commit 27d8ccae authored by Rakesh MEHTA's avatar Rakesh MEHTA
Browse files

Restructured machines and trainers

parent 26c6da62
Branches
Tags
No related merge requests found
Hide whitespace changes
Inline Side-by-side
xbob/boosting/core/boosting.py
+ 13
− 10
View file @ 27d8ccae
......@@ -26,7 +26,7 @@ Loss Function: Exponential Loss (Preferred with the StumpTrainer)
import numpy
import trainers
import losses
from scipy import optimize
import scipy.optimize
......@@ -160,15 +160,18 @@ class Boost:
#if(self.weak_trainer_type == 'LutTrainer'):
# fset = fset.astype(int)
# For each round of boosting initialize a new weak trainer
if(self.weak_trainer_type == 'LutTrainer'):
weak_trainer = trainers.LutTrainer(self.num_entries, self.lut_selection, num_op )
elif (self.weak_trainer_type == 'StumpTrainer'):
weak_trainer = trainers.StumpTrainer()
# Start boosting iterations for num_rnds rounds
for r in range(self.num_rnds):
# For each round of boosting initialize a new weak trainer
if(self.weak_trainer_type == 'LutTrainer'):
weak_trainer = trainers.LutTrainer(self.num_entries, self.lut_selection, num_op )
elif (self.weak_trainer_type == 'StumpTrainer'):
weak_trainer = trainers.StumpTrainer()
# Compute the gradient of the loss function, l'(y,f(x)) using loss_ class
loss_grad = loss_func.update_loss_grad(targets,pred_scores)
......@@ -176,14 +179,14 @@ class Boost:
curr_weak_trainer = weak_trainer.compute_weak_trainer(fset, loss_grad)
# Compute the classification scores of the samples based only on the current round weak classifier (g_r)
curr_pred_scores = weak_trainer.get_weak_scores(fset)
curr_pred_scores = curr_weak_trainer.get_weak_scores(fset)
# Initialize the start point for lbfgs minimization
f0 = numpy.zeros(num_op)
init_point = numpy.zeros(num_op)
# Perform lbfgs minimization and compute the scale (alpha_r) for current weak trainer
lbfgs_struct = optimize.fmin_l_bfgs_b(loss_func.loss_sum, f0, fprime = loss_func.loss_grad_sum, args = (targets, pred_scores, curr_pred_scores))
lbfgs_struct = scipy.optimize.fmin_l_bfgs_b(loss_func.loss_sum, init_point, fprime = loss_func.loss_grad_sum, args = (targets, pred_scores, curr_pred_scores))
alpha = lbfgs_struct[0]
......
xbob/boosting/core/trainers.py
+ 87
− 52
View file @ 27d8ccae
......@@ -13,17 +13,48 @@
import numpy
import math
class StumpMachine():
""" The StumpMachine class consist of the core elements of the Stump weak classfier i.e. the threshold,
the polarity and the feature index corresponding to the weak classifier. """
class StumpTrainer():
""" The weak trainer class for training stumps as classifiers. The trainer is parametrized
the threshold and the polarity.
"""
def __init__(self):
""" Initialize the stump classifier"""
self.threshold = 0
self.polarity = 0
self.selected_indices = 0
def get_weak_scores(self,test_features):
""" The function to perform classification using a weak stump classifier.
The function computes the classification scores for the test features using
a weak stump trainer. Since we use the stump classifier the classification
scores are either +1 or -1.
Input: self: a weak stump trainer
test_features: A matrix of the test features of dimension.
Num. of Test images x Num. of features
Return: weak_scores: classification scores of the test features use the weak classifier self
Array of dimension = Num. of samples
"""
# Initialize the values
numSamp = test_features.shape[0]
weak_scores = numpy.ones([numSamp,1])
# Select feature corresponding to the specific index
weak_features = test_features[:,self.selected_indices]
# classify the features and compute the score
weak_scores[weak_features < self.threshold] = -1
weak_scores = self.polarity *weak_scores
return weak_scores
class StumpTrainer():
""" The weak trainer class for training stumps as classifiers. The trainer is parametrized
the threshold and the polarity.
"""
......@@ -53,6 +84,7 @@ class StumpTrainer():
threshold = numpy.zeros([numFea])
polarity = numpy.zeros([numFea])
gain = numpy.zeros([numFea])
stump_machine = StumpMachine()
# For each feature find the optimum threshold, polarity and the gain
for i in range(numFea):
......@@ -60,10 +92,10 @@ class StumpTrainer():
# Find the optimum id and its corresponding trainer
opt_id = gain.argmax()
self.threshold = threshold[opt_id]
self.polarity = polarity[opt_id]
self.selected_indices = opt_id
return self
stump_machine.threshold = threshold[opt_id]
stump_machine.polarity = polarity[opt_id]
stump_machine.selected_indices = opt_id
return stump_machine
......@@ -123,33 +155,52 @@ class StumpTrainer():
def get_weak_scores(self,test_features):
""" The function to perform classification using a weak stump classifier.
The function computes the classification scores for the test features using
a weak stump trainer. Since we use the stump classifier the classification
scores are either +1 or -1.
Input: self: a weak stump trainer
test_features: A matrix of the test features of dimension.
Num. of Test images x Num. of features
Return: weak_scores: classification scores of the test features use the weak classifier self
Array of dimension = Num. of samples
class LutMachine():
""" The LUT machine consist of the core elements of the LUT weak classfier i.e. the LUT and
the feature index corresponding to the weak classifier. """
def __init__(self, num_outputs, num_entries):
""" The function initializes the weak LUT machine.
The function initializes the look-up-table and the feature indices of the LUT machine.
Inputs:
self:
num_entries: The number of entries for the LUT
type: int
num_outputs: The number of outputs for the classification task.
type: Integer
"""
# Initialize the values
numSamp = test_features.shape[0]
weak_scores = numpy.ones([numSamp,1])
# Select feature corresponding to the specific index
weak_features = test_features[:,self.selected_indices]
self.luts = numpy.ones((num_entries, num_outputs), dtype = numpy.int)
self.selected_indices = numpy.zeros([num_outputs,1], 'int16')
# classify the features and compute the score
weak_scores[weak_features < self.threshold] = -1
weak_scores = self.polarity *weak_scores
return weak_scores
def get_weak_scores(self, fset):
""" Function computes classification results according to the LUT machine
Function classifies the features based on a single LUT machine.
Input:
fset: The set test features. No. of test samples x No. of total features
return:
weak_scores: The classification scores of the features based on current weak classifier"""
# Initialize
num_samp = len(fset)
num_outputs = len(self.luts[0])
weak_scores = numpy.zeros([num_samp,num_outputs])
# Compute weak scores
for oi in range(num_outputs):
weak_scores[:,oi] = numpy.transpose(self.luts[fset[:,self.selected_indices[oi]],oi])
return weak_scores
class LutTrainer():
......@@ -181,6 +232,7 @@ class LutTrainer():
"""
self.num_entries = num_entries
self.num_outputs = num_outputs
self.luts = numpy.ones((num_entries, num_outputs), dtype = numpy.int)
self.selection_type = selection_type
self.selected_indices = numpy.zeros([num_outputs,1], 'int16')
......@@ -210,8 +262,8 @@ class LutTrainer():
# Initializations
num_outputs = loss_grad.shape[1]
print num_outputs
fea_grad = numpy.zeros([self.num_entries,num_outputs])
fea_grad = numpy.zeros([self.num_entries, num_outputs])
lut_machine = LutMachine(self.num_outputs, self.num_entries)
# Compute the sum of the gradient based on the feature values or the loss associated with each
# feature index
......@@ -231,8 +283,7 @@ class LutTrainer():
for oi in range(num_outputs):
curr_id = sum_loss[:,oi].argmin()
fea_grad[:,oi] = self.compute_grad_hist(loss_grad[:,oi],fea[:,curr_id])
print oi
self.selected_indices[oi] = curr_id
lut_machine.selected_indices[oi] = curr_id
elif self.selection_type == 'shared':
......@@ -242,14 +293,14 @@ class LutTrainer():
accum_loss = numpy.sum(sum_loss,1)
selected_findex = accum_loss.argmin()
self.selected_indices = selected_findex*numpy.ones([num_outputs,1],'int16')
lut_machine.selected_indices = selected_findex*numpy.ones([num_outputs,1],'int16')
for oi in range(num_outputs):
fea_grad[:,oi] = self.compute_grad_hist(loss_grad[:,oi],fea[:,selected_findex])
# Assign the values to LookUp Table
self.luts[fea_grad <= 0.0] = -1
return self
lut_machine.luts[fea_grad <= 0.0] = -1
return lut_machine
......@@ -310,21 +361,5 @@ class LutTrainer():
def get_weak_scores(self, fset):
""" Function computes classification results according to current weak classifier
Function classifies the features based on a single weak classifier.
Input:
fset: The set test features. No. of test samples x No. of total features
return:
weak_scores: The classification scores of the features based on current weak classifier"""
num_samp = len(fset)
num_outputs = len(self.luts[0])
weak_scores = numpy.zeros([num_samp,num_outputs])
for oi in range(num_outputs):
a = self.luts[fset[:,self.selected_indices[oi]],oi]
weak_scores[:,oi] = numpy.transpose(self.luts[fset[:,self.selected_indices[oi]],oi])
return weak_scores
0% or .
You are about to add 0 people to the discussion. Proceed with caution.
Please register or to comment