From afce6631fdd2785c3ffb831fa75ba7e1b0bab2ea Mon Sep 17 00:00:00 2001
From: Thomas Weber <Thomas.Weber@physik.uni-muenchen.de>
Date: Fri, 4 May 2018 16:42:06 +0200
Subject: [PATCH] Added new genetic algorithm, might be better
https://github.com/jliphard/DeepEvolve
---
.gitignore | 1 +
Genetic_Algorithm/allgenomes.py | 62 +++
Genetic_Algorithm/evolver.py | 259 ++++++++++
Genetic_Algorithm/genome.py | 132 ++++++
Genetic_Algorithm/idgen.py | 31 ++
Genetic_Algorithm/log.txt | 574 -----------------------
Genetic_Algorithm/main.py | 131 +++---
Genetic_Algorithm/old/main.py | 119 +++++
Genetic_Algorithm/{ => old}/network.py | 0
Genetic_Algorithm/{ => old}/optimizer.py | 0
Genetic_Algorithm/old/train.py | 55 +++
Genetic_Algorithm/train.py | 34 +-
toolkit.py | 28 +-
13 files changed, 772 insertions(+), 654 deletions(-)
create mode 100755 Genetic_Algorithm/allgenomes.py
create mode 100755 Genetic_Algorithm/evolver.py
create mode 100755 Genetic_Algorithm/genome.py
create mode 100755 Genetic_Algorithm/idgen.py
delete mode 100755 Genetic_Algorithm/log.txt
create mode 100755 Genetic_Algorithm/old/main.py
rename Genetic_Algorithm/{ => old}/network.py (100%)
rename Genetic_Algorithm/{ => old}/optimizer.py (100%)
create mode 100755 Genetic_Algorithm/old/train.py
diff --git a/.gitignore b/.gitignore
index 38538a2..4921c9e 100644
--- a/.gitignore
+++ b/.gitignore
@@ -4,3 +4,4 @@ run.py
*.swp
*.pyc
*.pdf
+*.txt
diff --git a/Genetic_Algorithm/allgenomes.py b/Genetic_Algorithm/allgenomes.py
new file mode 100755
index 0000000..54eb0ad
--- /dev/null
+++ b/Genetic_Algorithm/allgenomes.py
@@ -0,0 +1,62 @@
+#!/usr/bin/env python
+
+""" Class that keeps track of all genomes trained so far, and their scores.
+ Among other things, ensures that genomes are unique.
+"""
+
+import random
+import logging
+
+from genome import Genome
+
+class AllGenomes(object):
+ """Store all genomes
+ """
+
+ def __init__(self, firstgenome):
+ """Initialize
+ """
+
+ self.population = []
+ self.population.append(firstgenome)
+
+ def add_genome(self, genome):
+ """Add the genome to our population.
+ """
+
+ for i in range(0,len(self.population)):
+ if (genome.hash == self.population[i].hash):
+ logging.info("add_genome() ERROR: hash clash - duplicate genome")
+ return False
+
+ self.population.append(genome)
+
+ return True
+
+ def set_accuracy(self, genome):
+ """Add the genome to our population.
+ """
+
+ for i in range(0,len(self.population)):
+ if (genome.hash == self.population[i].hash):
+ self.population[i].accuracy = genome.accuracy
+ return
+
+ logging.info("set_accuracy() ERROR: Genome not found")
+
+ def is_duplicate(self, genome):
+ """Add the genome to our population.
+ """
+
+ for i in range(0,len(self.population)):
+ if (genome.hash == self.population[i].hash):
+ return True
+
+ return False
+
+ def print_all_genomes(self):
+ """Print out a genome.
+ """
+
+ for genome in self.population:
+ genome.print_genome_ma()
diff --git a/Genetic_Algorithm/evolver.py b/Genetic_Algorithm/evolver.py
new file mode 100755
index 0000000..5ab802b
--- /dev/null
+++ b/Genetic_Algorithm/evolver.py
@@ -0,0 +1,259 @@
+#!/usr/bin/env python
+
+"""
+Class that holds a genetic algorithm for evolving a network.
+
+Inspiration:
+
+ http://lethain.com/genetic-algorithms-cool-name-damn-simple/
+"""
+from __future__ import print_function
+
+import random
+import logging
+import copy
+
+from functools import reduce
+from operator import add
+from genome import Genome
+from idgen import IDgen
+from allgenomes import AllGenomes
+
+
+class Evolver(object):
+ """Class that implements genetic algorithm."""
+
+ def __init__(self, all_possible_genes, retain=0.15, random_select=0.1, mutate_chance=0.3):
+ """Create an optimizer.
+
+ Args:
+ all_possible_genes (dict): Possible genome parameters
+ retain (float): Percentage of population to retain after
+ each generation
+ random_select (float): Probability of a rejected genome
+ remaining in the population
+ mutate_chance (float): Probability a genome will be
+ randomly mutated
+
+ """
+
+ self.all_possible_genes = all_possible_genes
+ self.retain = retain
+ self.random_select = random_select
+ self.mutate_chance = mutate_chance
+
+ #set the ID gen
+ self.ids = IDgen()
+
+ def create_population(self, count):
+ """Create a population of random networks.
+
+ Args:
+ count (int): Number of networks to generate, aka the
+ size of the population
+
+ Returns:
+ (list): Population of network objects
+
+ """
+ pop = []
+
+ i = 0
+
+ while i < count:
+
+ # Initialize a new genome.
+ genome = Genome( self.all_possible_genes, {}, self.ids.get_next_ID(), 0, 0, self.ids.get_Gen() )
+
+ # Set it to random parameters.
+ genome.set_genes_random()
+
+ if i == 0:
+ #this is where we will store all genomes
+ self.master = AllGenomes( genome )
+ else:
+ # Make sure it is unique....
+ while self.master.is_duplicate( genome ):
+ genome.mutate_one_gene()
+
+ # Add the genome to our population.
+ pop.append(genome)
+
+ # and add to the master list
+ if i > 0:
+ self.master.add_genome(genome)
+
+ i += 1
+
+ #self.master.print_all_genomes()
+
+ #exit()
+
+ return pop
+
+ @staticmethod
+ def fitness(genome):
+ """Return the accuracy, which is our fitness function."""
+ return genome.accuracy
+
+ def grade(self, pop):
+ """Find average fitness for a population.
+
+ Args:
+ pop (list): The population of networks/genome
+
+ Returns:
+ (float): The average accuracy of the population
+
+ """
+ summed = reduce(add, (self.fitness(genome) for genome in pop))
+ return summed / float((len(pop)))
+
+ def breed(self, mom, dad):
+ """Make two children from parental genes.
+
+ Args:
+ mother (dict): genome parameters
+ father (dict): genome parameters
+
+ Returns:
+ (list): Two network objects
+
+ """
+ children = []
+
+ #where do we recombine? 0, 1, 2, 3, 4... N?
+ #with four genes, there are three choices for the recombination
+ # ___ * ___ * ___ * ___
+ #0 -> no recombination, and N == length of dictionary -> no recombination
+ #0 and 4 just (re)create more copies of the parents
+ #so the range is always 1 to len(all_possible_genes) - 1
+ pcl = len(self.all_possible_genes)
+
+ recomb_loc = random.randint(1,pcl - 1)
+
+ #for _ in range(2): #make _two_ children - could also make more
+ child1 = {}
+ child2 = {}
+
+ #enforce defined genome order using list
+ #keys = ['nb_neurons', 'nb_layers', 'activation', 'optimizer']
+ keys = list(self.all_possible_genes)
+ keys = sorted(keys) #paranoia - just to make sure we do not add unintentional randomization
+
+ #*** CORE RECOMBINATION CODE ****
+ for x in range(0, pcl):
+ if x < recomb_loc:
+ child1[keys[x]] = mom.geneparam[keys[x]]
+ child2[keys[x]] = dad.geneparam[keys[x]]
+ else:
+ child1[keys[x]] = dad.geneparam[keys[x]]
+ child2[keys[x]] = mom.geneparam[keys[x]]
+
+ # Initialize a new genome
+ # Set its parameters to those just determined
+ # they both have the same mom and dad
+ genome1 = Genome( self.all_possible_genes, child1, self.ids.get_next_ID(), mom.u_ID, dad.u_ID, self.ids.get_Gen() )
+ genome2 = Genome( self.all_possible_genes, child2, self.ids.get_next_ID(), mom.u_ID, dad.u_ID, self.ids.get_Gen() )
+
+ #at this point, there is zero guarantee that the genome is actually unique
+
+ # Randomly mutate one gene
+ if self.mutate_chance > random.random():
+ genome1.mutate_one_gene()
+
+ if self.mutate_chance > random.random():
+ genome2.mutate_one_gene()
+
+ #do we have a unique child or are we just retraining one we already have anyway?
+ while self.master.is_duplicate(genome1):
+ genome1.mutate_one_gene()
+
+ self.master.add_genome(genome1)
+
+ while self.master.is_duplicate(genome2):
+ genome2.mutate_one_gene()
+
+ self.master.add_genome(genome2)
+
+ children.append(genome1)
+ children.append(genome2)
+
+ return children
+
+ def evolve(self, pop):
+ """Evolve a population of genomes.
+
+ Args:
+ pop (list): A list of genome parameters
+
+ Returns:
+ (list): The evolved population of networks
+
+ """
+ #increase generation
+ self.ids.increase_Gen()
+
+ # Get scores for each genome
+ graded = [(self.fitness(genome), genome) for genome in pop]
+
+ #and use those scores to fill in the master list
+ for genome in pop:
+ self.master.set_accuracy(genome)
+
+ # Sort on the scores.
+ graded = [x[1] for x in sorted(graded, key=lambda x: x[0], reverse=True)]
+
+ # Get the number we want to keep unchanged for the next cycle.
+ retain_length = int(len(graded)*self.retain)
+
+ # In this first step, we keep the 'top' X percent (as defined in self.retain)
+ # We will not change them, except we will update the generation
+ new_generation = graded[:retain_length]
+
+ # For the lower scoring ones, randomly keep some anyway.
+ # This is wasteful, since we _know_ these are bad, so why keep rescoring them without modification?
+ # At least we should mutate them
+ for genome in graded[retain_length:]:
+ if self.random_select > random.random():
+ gtc = copy.deepcopy(genome)
+
+ while self.master.is_duplicate(gtc):
+ gtc.mutate_one_gene()
+
+ gtc.set_generation( self.ids.get_Gen() )
+ new_generation.append(gtc)
+ self.master.add_genome(gtc)
+
+ # Now find out how many spots we have left to fill.
+ ng_length = len(new_generation)
+
+ desired_length = len(pop) - ng_length
+
+ children = []
+
+ # Add children, which are bred from pairs of remaining (i.e. very high or lower scoring) genomes.
+ while len(children) < desired_length:
+
+ # Get a random mom and dad, but, need to make sure they are distinct
+ parents = random.sample(range(ng_length-1), k=2)
+
+ i_male = parents[0]
+ i_female = parents[1]
+
+ male = new_generation[i_male]
+ female = new_generation[i_female]
+
+ # Recombine and mutate
+ babies = self.breed(male, female)
+ # the babies are guaranteed to be novel
+
+ # Add the children one at a time.
+ for baby in babies:
+ # Don't grow larger than desired length.
+ #if len(children) < desired_length:
+ children.append(baby)
+
+ new_generation.extend(children)
+
+ return new_generation
diff --git a/Genetic_Algorithm/genome.py b/Genetic_Algorithm/genome.py
new file mode 100755
index 0000000..4c5dc25
--- /dev/null
+++ b/Genetic_Algorithm/genome.py
@@ -0,0 +1,132 @@
+#!/usr/bin/env python
+
+"""The genome to be evolved."""
+
+import random
+import logging
+import hashlib
+import copy
+
+from train import train_and_score
+
+class Genome(object):
+ """
+ Represents one genome and all relevant utility functions (add, mutate, etc.).
+ """
+
+ def __init__( self, all_possible_genes = None, geneparam = {}, u_ID = 0, mom_ID = 0, dad_ID = 0, gen = 0 ):
+ """Initialize a genome.
+
+ Args:
+ all_possible_genes (dict): Parameters for the genome, includes:
+ gene_nb_neurons (list): [64, 128, 256]
+ gene_nb_layers (list): [1, 2, 3, 4]
+ gene_activation (list): ['relu', 'elu']
+ gene_optimizer (list): ['rmsprop', 'adam']
+ """
+ self.accuracy = 0.0
+ self.all_possible_genes = all_possible_genes
+ self.geneparam = geneparam #(dict): represents actual genome parameters
+ self.u_ID = u_ID
+ self.parents = [mom_ID, dad_ID]
+ self.generation = gen
+
+ #hash only makes sense when we have specified the genes
+ if not geneparam:
+ self.hash = 0
+ else:
+ self.update_hash()
+
+ def update_hash(self):
+ """
+ Refesh each genome's unique hash - needs to run after any genome changes.
+ """
+ # + str(self.geneparam['optimizer'])
+ genh = str(self.geneparam['nb_neurons']) + self.geneparam['activation'] \
+ + str(self.geneparam['nb_layers']) \
+ + str(self.geneparam['lr']) \
+ + str(self.geneparam['decay']) \
+ + str(self.geneparam['momentum'])
+
+ self.hash = hashlib.md5(genh.encode("UTF-8")).hexdigest()
+
+ self.accuracy = 0.0
+
+ def set_genes_random(self):
+ """Create a random genome."""
+ #print("set_genes_random")
+ self.parents = [0,0] #very sad - no parents :(
+
+ for key in self.all_possible_genes:
+ self.geneparam[key] = random.choice(self.all_possible_genes[key])
+
+ self.update_hash()
+
+ def mutate_one_gene(self):
+ """Randomly mutate one gene in the genome.
+
+ Args:
+ network (dict): The genome parameters to mutate
+
+ Returns:
+ (Genome): A randomly mutated genome object
+
+ """
+ # Which gene shall we mutate? Choose one of N possible keys/genes.
+ gene_to_mutate = random.choice( list(self.all_possible_genes.keys()) )
+
+ # And then let's mutate one of the genes.
+ # Make sure that this actually creates mutation
+ current_value = self.geneparam[gene_to_mutate]
+ possible_choices = copy.deepcopy(self.all_possible_genes[gene_to_mutate])
+
+ possible_choices.remove(current_value)
+
+ self.geneparam[gene_to_mutate] = random.choice( possible_choices )
+
+ self.update_hash()
+
+ def set_generation(self, generation):
+ """needed when a genome is passed on from one generation to the next.
+ the id stays the same, but the generation is increased"""
+
+ self.generation = generation
+ #logging.info("Setting Generation to %d" % self.generation)
+
+ def set_genes_to(self, geneparam, mom_ID, dad_ID):
+ """Set genome properties.
+ this is used when breeding kids
+
+ Args:
+ genome (dict): The genome parameters
+ IMPROVE
+ """
+ self.parents = [mom_ID, dad_ID]
+
+ self.geneparam = geneparam
+
+ self.update_hash()
+
+ def train(self):
+ """Train the genome and record the accuracy.
+
+ Args:
+
+ """
+ if self.accuracy == 0.0: #don't bother retraining ones we already trained
+ self.accuracy = train_and_score(self.geneparam)
+
+ def print_genome(self):
+ """Print out a genome."""
+ logging.info(self.geneparam)
+ logging.info("Acc: %.2f%%" % (self.accuracy * 100))
+ logging.info("UniID: %d" % self.u_ID)
+ logging.info("Mom and Dad: %d %d" % (self.parents[0], self.parents[1]))
+ logging.info("Gen: %d" % self.generation)
+ logging.info("Hash: %s" % self.hash)
+
+ def print_genome_ma(self):
+ """Print out a genome."""
+ logging.info(self.geneparam)
+ logging.info("Acc: %.2f%% UniID: %d Mom and Dad: %d %d Gen: %d" % (self.accuracy * 100, self.u_ID, self.parents[0], self.parents[1], self.generation))
+ logging.info("Hash: %s" % self.hash)
diff --git a/Genetic_Algorithm/idgen.py b/Genetic_Algorithm/idgen.py
new file mode 100755
index 0000000..5c2a09b
--- /dev/null
+++ b/Genetic_Algorithm/idgen.py
@@ -0,0 +1,31 @@
+#!/usr/bin/env python
+
+"""Provide unique genome IDs."""
+
+import logging
+
+class IDgen():
+ """Generate unique IDs.
+ """
+
+ def __init__(self):
+ """Keep track of IDs.
+ """
+ self.currentID = 0
+ self.currentGen = 1
+
+ def get_next_ID(self):
+
+ self.currentID += 1
+
+ return self.currentID
+
+ def increase_Gen(self):
+
+ self.currentGen += 1
+
+ def get_Gen(self):
+
+ return self.currentGen
+
+
diff --git a/Genetic_Algorithm/log.txt b/Genetic_Algorithm/log.txt
deleted file mode 100755
index fed7212..0000000
--- a/Genetic_Algorithm/log.txt
+++ /dev/null
@@ -1,574 +0,0 @@
-05/04/2018 12:06:59 PM - INFO - ***Evolving 7 generations with population 5***
-05/04/2018 12:06:59 PM - INFO - ***Doing generation 1 of 7***
-05/04/2018 12:06:59 PM - INFO - Trying to load x_train from ./outputs/x_train.h5
-05/04/2018 12:06:59 PM - INFO - Couldn't load all datasets - reading from ROOT trees
-05/04/2018 12:07:01 PM - DEBUG - training data before transformation: [[369.8166 236.10608 ]
- [217.16876 348.09952 ]
- [395.43085 142.5729 ]
- ...
- [223.76315 101.57288 ]
- [561.4382 47.980972]
- [200.53334 70.21298 ]]
-05/04/2018 12:07:01 PM - DEBUG - minimum values: [200.00388, 0.00040848268]
-05/04/2018 12:07:01 PM - DEBUG - maximum values: [5745.249, 6567.212]
-05/04/2018 12:07:01 PM - INFO - Creating new RobustScaler
-05/04/2018 12:07:01 PM - INFO - Fitting RobustScaler to training data
-05/04/2018 12:07:01 PM - DEBUG - training data after transformation: [[ 0.7659683 3.400636 ]
- [-0.44083464 5.5718575 ]
- [ 0.9684695 1.5873041 ]
- ...
- [-0.38870072 0.79243493]
- [ 2.28089 -0.24655382]
- [-0.5723509 0.18445913]]
-05/04/2018 12:07:01 PM - INFO - Training and test data transformed
-05/04/2018 12:07:01 PM - INFO - Train model
-05/04/2018 12:07:01 PM - INFO - Using SGD(**{'lr': 0.5, 'momentum': 0.7, 'decay': 1e-05}) as Optimizer
-05/04/2018 12:07:01 PM - INFO - Compile model
-05/04/2018 12:07:01 PM - INFO - No weights found, starting completely new model
-05/04/2018 12:07:01 PM - DEBUG - Calculated class_weight: [0.5045154119594469, 55.865933882725024]
-05/04/2018 12:07:01 PM - WARNING - Variable *= will be deprecated. Use variable.assign_mul if you want assignment to the variable value or 'x = x * y' if you want a new python Tensor object.
-05/04/2018 12:08:15 PM - INFO - Get test loss and metrics of the model
-05/04/2018 12:08:18 PM - INFO - Trying to load x_train from ./outputs/x_train.h5
-05/04/2018 12:08:18 PM - INFO - Couldn't load all datasets - reading from ROOT trees
-05/04/2018 12:08:20 PM - DEBUG - training data before transformation: [[369.8166 236.10608 ]
- [217.16876 348.09952 ]
- [395.43085 142.5729 ]
- ...
- [223.76315 101.57288 ]
- [561.4382 47.980972]
- [200.53334 70.21298 ]]
-05/04/2018 12:08:20 PM - DEBUG - minimum values: [200.00388, 0.00040848268]
-05/04/2018 12:08:20 PM - DEBUG - maximum values: [5745.249, 6567.212]
-05/04/2018 12:08:20 PM - INFO - Creating new RobustScaler
-05/04/2018 12:08:20 PM - INFO - Fitting RobustScaler to training data
-05/04/2018 12:08:20 PM - DEBUG - training data after transformation: [[ 0.7659683 3.400636 ]
- [-0.44083464 5.5718575 ]
- [ 0.9684695 1.5873041 ]
- ...
- [-0.38870072 0.79243493]
- [ 2.28089 -0.24655382]
- [-0.5723509 0.18445913]]
-05/04/2018 12:08:20 PM - INFO - Training and test data transformed
-05/04/2018 12:08:20 PM - INFO - Train model
-05/04/2018 12:08:20 PM - INFO - Using SGD(**{'lr': 1.0, 'momentum': 0.0, 'decay': 0.1}) as Optimizer
-05/04/2018 12:08:20 PM - INFO - Compile model
-05/04/2018 12:08:20 PM - INFO - No weights found, starting completely new model
-05/04/2018 12:08:20 PM - DEBUG - Calculated class_weight: [0.5045154119594469, 55.865933882725024]
-05/04/2018 12:18:00 PM - INFO - Get test loss and metrics of the model
-05/04/2018 12:18:12 PM - INFO - Trying to load x_train from ./outputs/x_train.h5
-05/04/2018 12:18:12 PM - INFO - Couldn't load all datasets - reading from ROOT trees
-05/04/2018 12:18:14 PM - DEBUG - training data before transformation: [[369.8166 236.10608 ]
- [217.16876 348.09952 ]
- [395.43085 142.5729 ]
- ...
- [223.76315 101.57288 ]
- [561.4382 47.980972]
- [200.53334 70.21298 ]]
-05/04/2018 12:18:14 PM - DEBUG - minimum values: [200.00388, 0.00040848268]
-05/04/2018 12:18:14 PM - DEBUG - maximum values: [5745.249, 6567.212]
-05/04/2018 12:18:14 PM - INFO - Creating new RobustScaler
-05/04/2018 12:18:14 PM - INFO - Fitting RobustScaler to training data
-05/04/2018 12:18:14 PM - DEBUG - training data after transformation: [[ 0.7659683 3.400636 ]
- [-0.44083464 5.5718575 ]
- [ 0.9684695 1.5873041 ]
- ...
- [-0.38870072 0.79243493]
- [ 2.28089 -0.24655382]
- [-0.5723509 0.18445913]]
-05/04/2018 12:18:14 PM - INFO - Training and test data transformed
-05/04/2018 12:18:14 PM - INFO - Train model
-05/04/2018 12:18:14 PM - INFO - Using SGD(**{'lr': 0.5, 'momentum': 0.5, 'decay': 0.01}) as Optimizer
-05/04/2018 12:18:14 PM - INFO - Compile model
-05/04/2018 12:18:14 PM - INFO - No weights found, starting completely new model
-05/04/2018 12:18:14 PM - DEBUG - Calculated class_weight: [0.5045154119594469, 55.86593388272502]
-05/04/2018 12:18:24 PM - INFO - Get test loss and metrics of the model
-05/04/2018 12:18:25 PM - INFO - Trying to load x_train from ./outputs/x_train.h5
-05/04/2018 12:18:25 PM - INFO - Couldn't load all datasets - reading from ROOT trees
-05/04/2018 12:18:26 PM - DEBUG - training data before transformation: [[369.8166 236.10608 ]
- [217.16876 348.09952 ]
- [395.43085 142.5729 ]
- ...
- [223.76315 101.57288 ]
- [561.4382 47.980972]
- [200.53334 70.21298 ]]
-05/04/2018 12:18:26 PM - DEBUG - minimum values: [200.00388, 0.00040848268]
-05/04/2018 12:18:26 PM - DEBUG - maximum values: [5745.249, 6567.212]
-05/04/2018 12:18:26 PM - INFO - Creating new RobustScaler
-05/04/2018 12:18:26 PM - INFO - Fitting RobustScaler to training data
-05/04/2018 12:18:26 PM - DEBUG - training data after transformation: [[ 0.7659683 3.400636 ]
- [-0.44083464 5.5718575 ]
- [ 0.9684695 1.5873041 ]
- ...
- [-0.38870072 0.79243493]
- [ 2.28089 -0.24655382]
- [-0.5723509 0.18445913]]
-05/04/2018 12:18:26 PM - INFO - Training and test data transformed
-05/04/2018 12:18:26 PM - INFO - Train model
-05/04/2018 12:18:26 PM - INFO - Using SGD(**{'lr': 0.1, 'momentum': 0.1, 'decay': 1e-07}) as Optimizer
-05/04/2018 12:18:26 PM - INFO - Compile model
-05/04/2018 12:18:26 PM - INFO - No weights found, starting completely new model
-05/04/2018 12:18:26 PM - DEBUG - Calculated class_weight: [0.5045154119594469, 55.86593388272504]
-05/04/2018 12:23:55 PM - INFO - Get test loss and metrics of the model
-05/04/2018 12:24:03 PM - INFO - Trying to load x_train from ./outputs/x_train.h5
-05/04/2018 12:24:03 PM - INFO - Couldn't load all datasets - reading from ROOT trees
-05/04/2018 12:24:05 PM - DEBUG - training data before transformation: [[369.8166 236.10608 ]
- [217.16876 348.09952 ]
- [395.43085 142.5729 ]
- ...
- [223.76315 101.57288 ]
- [561.4382 47.980972]
- [200.53334 70.21298 ]]
-05/04/2018 12:24:05 PM - DEBUG - minimum values: [200.00388, 0.00040848268]
-05/04/2018 12:24:05 PM - DEBUG - maximum values: [5745.249, 6567.212]
-05/04/2018 12:24:05 PM - INFO - Creating new RobustScaler
-05/04/2018 12:24:05 PM - INFO - Fitting RobustScaler to training data
-05/04/2018 12:24:05 PM - DEBUG - training data after transformation: [[ 0.7659683 3.400636 ]
- [-0.44083464 5.5718575 ]
- [ 0.9684695 1.5873041 ]
- ...
- [-0.38870072 0.79243493]
- [ 2.28089 -0.24655382]
- [-0.5723509 0.18445913]]
-05/04/2018 12:24:05 PM - INFO - Training and test data transformed
-05/04/2018 12:24:05 PM - INFO - Train model
-05/04/2018 12:24:05 PM - INFO - Using SGD(**{'lr': 0.1, 'momentum': 1.0, 'decay': 0.01}) as Optimizer
-05/04/2018 12:24:05 PM - INFO - Compile model
-05/04/2018 12:24:05 PM - INFO - No weights found, starting completely new model
-05/04/2018 12:24:05 PM - DEBUG - Calculated class_weight: [0.5045154119594469, 55.865933882725024]
-05/04/2018 12:26:31 PM - INFO - Get test loss and metrics of the model
-05/04/2018 12:26:35 PM - INFO - Generation average: 58.25%
-05/04/2018 12:26:35 PM - INFO - --------------------------------------------------------------------------------
-05/04/2018 12:26:35 PM - INFO - ***Doing generation 2 of 7***
-05/04/2018 12:26:35 PM - INFO - Trying to load x_train from ./outputs/x_train.h5
-05/04/2018 12:26:35 PM - INFO - Couldn't load all datasets - reading from ROOT trees
-05/04/2018 12:26:37 PM - DEBUG - training data before transformation: [[369.8166 236.10608 ]
- [217.16876 348.09952 ]
- [395.43085 142.5729 ]
- ...
- [223.76315 101.57288 ]
- [561.4382 47.980972]
- [200.53334 70.21298 ]]
-05/04/2018 12:26:37 PM - DEBUG - minimum values: [200.00388, 0.00040848268]
-05/04/2018 12:26:37 PM - DEBUG - maximum values: [5745.249, 6567.212]
-05/04/2018 12:26:37 PM - INFO - Creating new RobustScaler
-05/04/2018 12:26:37 PM - INFO - Fitting RobustScaler to training data
-05/04/2018 12:26:37 PM - DEBUG - training data after transformation: [[ 0.7659683 3.400636 ]
- [-0.44083464 5.5718575 ]
- [ 0.9684695 1.5873041 ]
- ...
- [-0.38870072 0.79243493]
- [ 2.28089 -0.24655382]
- [-0.5723509 0.18445913]]
-05/04/2018 12:26:37 PM - INFO - Training and test data transformed
-05/04/2018 12:26:37 PM - INFO - Train model
-05/04/2018 12:26:37 PM - INFO - Using SGD(**{'lr': 0.5, 'momentum': 1.0, 'decay': 1e-05}) as Optimizer
-05/04/2018 12:26:37 PM - INFO - Compile model
-05/04/2018 12:26:37 PM - INFO - No weights found, starting completely new model
-05/04/2018 12:26:37 PM - DEBUG - Calculated class_weight: [0.5045154119594469, 55.865933882725024]
-05/04/2018 12:27:48 PM - INFO - Get test loss and metrics of the model
-05/04/2018 12:27:52 PM - INFO - Trying to load x_train from ./outputs/x_train.h5
-05/04/2018 12:27:52 PM - INFO - Couldn't load all datasets - reading from ROOT trees
-05/04/2018 12:27:54 PM - DEBUG - training data before transformation: [[369.8166 236.10608 ]
- [217.16876 348.09952 ]
- [395.43085 142.5729 ]
- ...
- [223.76315 101.57288 ]
- [561.4382 47.980972]
- [200.53334 70.21298 ]]
-05/04/2018 12:27:54 PM - DEBUG - minimum values: [200.00388, 0.00040848268]
-05/04/2018 12:27:54 PM - DEBUG - maximum values: [5745.249, 6567.212]
-05/04/2018 12:27:54 PM - INFO - Creating new RobustScaler
-05/04/2018 12:27:54 PM - INFO - Fitting RobustScaler to training data
-05/04/2018 12:27:54 PM - DEBUG - training data after transformation: [[ 0.7659683 3.400636 ]
- [-0.44083464 5.5718575 ]
- [ 0.9684695 1.5873041 ]
- ...
- [-0.38870072 0.79243493]
- [ 2.28089 -0.24655382]
- [-0.5723509 0.18445913]]
-05/04/2018 12:27:54 PM - INFO - Training and test data transformed
-05/04/2018 12:27:54 PM - INFO - Train model
-05/04/2018 12:27:54 PM - INFO - Using SGD(**{'lr': 0.1, 'momentum': 1.0, 'decay': 1e-05}) as Optimizer
-05/04/2018 12:27:54 PM - INFO - Compile model
-05/04/2018 12:27:54 PM - INFO - No weights found, starting completely new model
-05/04/2018 12:27:54 PM - DEBUG - Calculated class_weight: [0.5045154119594469, 55.86593388272504]
-05/04/2018 12:28:53 PM - INFO - Get test loss and metrics of the model
-05/04/2018 12:28:55 PM - INFO - Trying to load x_train from ./outputs/x_train.h5
-05/04/2018 12:28:55 PM - INFO - Couldn't load all datasets - reading from ROOT trees
-05/04/2018 12:28:57 PM - DEBUG - training data before transformation: [[369.8166 236.10608 ]
- [217.16876 348.09952 ]
- [395.43085 142.5729 ]
- ...
- [223.76315 101.57288 ]
- [561.4382 47.980972]
- [200.53334 70.21298 ]]
-05/04/2018 12:28:57 PM - DEBUG - minimum values: [200.00388, 0.00040848268]
-05/04/2018 12:28:57 PM - DEBUG - maximum values: [5745.249, 6567.212]
-05/04/2018 12:28:57 PM - INFO - Creating new RobustScaler
-05/04/2018 12:28:57 PM - INFO - Fitting RobustScaler to training data
-05/04/2018 12:28:57 PM - DEBUG - training data after transformation: [[ 0.7659683 3.400636 ]
- [-0.44083464 5.5718575 ]
- [ 0.9684695 1.5873041 ]
- ...
- [-0.38870072 0.79243493]
- [ 2.28089 -0.24655382]
- [-0.5723509 0.18445913]]
-05/04/2018 12:28:57 PM - INFO - Training and test data transformed
-05/04/2018 12:28:57 PM - INFO - Train model
-05/04/2018 12:28:57 PM - INFO - Using SGD(**{'lr': 0.5, 'momentum': 1.0, 'decay': 0.1}) as Optimizer
-05/04/2018 12:28:57 PM - INFO - Compile model
-05/04/2018 12:28:57 PM - INFO - No weights found, starting completely new model
-05/04/2018 12:28:57 PM - DEBUG - Calculated class_weight: [0.5045154119594469, 55.865933882725024]
-05/04/2018 12:31:28 PM - INFO - Get test loss and metrics of the model
-05/04/2018 12:31:32 PM - INFO - Generation average: 94.69%
-05/04/2018 12:31:32 PM - INFO - --------------------------------------------------------------------------------
-05/04/2018 12:31:32 PM - INFO - ***Doing generation 3 of 7***
-05/04/2018 12:31:32 PM - INFO - Trying to load x_train from ./outputs/x_train.h5
-05/04/2018 12:31:32 PM - INFO - Couldn't load all datasets - reading from ROOT trees
-05/04/2018 12:31:34 PM - DEBUG - training data before transformation: [[369.8166 236.10608 ]
- [217.16876 348.09952 ]
- [395.43085 142.5729 ]
- ...
- [223.76315 101.57288 ]
- [561.4382 47.980972]
- [200.53334 70.21298 ]]
-05/04/2018 12:31:34 PM - DEBUG - minimum values: [200.00388, 0.00040848268]
-05/04/2018 12:31:34 PM - DEBUG - maximum values: [5745.249, 6567.212]
-05/04/2018 12:31:34 PM - INFO - Creating new RobustScaler
-05/04/2018 12:31:34 PM - INFO - Fitting RobustScaler to training data
-05/04/2018 12:31:34 PM - DEBUG - training data after transformation: [[ 0.7659683 3.400636 ]
- [-0.44083464 5.5718575 ]
- [ 0.9684695 1.5873041 ]
- ...
- [-0.38870072 0.79243493]
- [ 2.28089 -0.24655382]
- [-0.5723509 0.18445913]]
-05/04/2018 12:31:34 PM - INFO - Training and test data transformed
-05/04/2018 12:31:34 PM - INFO - Train model
-05/04/2018 12:31:34 PM - INFO - Using SGD(**{'lr': 0.1, 'momentum': 1.0, 'decay': 0.01}) as Optimizer
-05/04/2018 12:31:34 PM - INFO - Compile model
-05/04/2018 12:31:34 PM - INFO - No weights found, starting completely new model
-05/04/2018 12:31:34 PM - DEBUG - Calculated class_weight: [0.5045154119594469, 55.865933882725024]
-05/04/2018 12:34:08 PM - INFO - Get test loss and metrics of the model
-05/04/2018 12:34:13 PM - INFO - Trying to load x_train from ./outputs/x_train.h5
-05/04/2018 12:34:13 PM - INFO - Couldn't load all datasets - reading from ROOT trees
-05/04/2018 12:34:14 PM - DEBUG - training data before transformation: [[369.8166 236.10608 ]
- [217.16876 348.09952 ]
- [395.43085 142.5729 ]
- ...
- [223.76315 101.57288 ]
- [561.4382 47.980972]
- [200.53334 70.21298 ]]
-05/04/2018 12:34:14 PM - DEBUG - minimum values: [200.00388, 0.00040848268]
-05/04/2018 12:34:14 PM - DEBUG - maximum values: [5745.249, 6567.212]
-05/04/2018 12:34:14 PM - INFO - Creating new RobustScaler
-05/04/2018 12:34:14 PM - INFO - Fitting RobustScaler to training data
-05/04/2018 12:34:14 PM - DEBUG - training data after transformation: [[ 0.7659683 3.400636 ]
- [-0.44083464 5.5718575 ]
- [ 0.9684695 1.5873041 ]
- ...
- [-0.38870072 0.79243493]
- [ 2.28089 -0.24655382]
- [-0.5723509 0.18445913]]
-05/04/2018 12:34:14 PM - INFO - Training and test data transformed
-05/04/2018 12:34:14 PM - INFO - Train model
-05/04/2018 12:34:14 PM - INFO - Using SGD(**{'lr': 0.5, 'momentum': 0.1, 'decay': 1e-05}) as Optimizer
-05/04/2018 12:34:14 PM - INFO - Compile model
-05/04/2018 12:34:14 PM - INFO - No weights found, starting completely new model
-05/04/2018 12:34:14 PM - DEBUG - Calculated class_weight: [0.5045154119594469, 55.86593388272502]
-05/04/2018 12:36:42 PM - INFO - Get test loss and metrics of the model
-05/04/2018 12:36:47 PM - INFO - Trying to load x_train from ./outputs/x_train.h5
-05/04/2018 12:36:47 PM - INFO - Couldn't load all datasets - reading from ROOT trees
-05/04/2018 12:36:49 PM - DEBUG - training data before transformation: [[369.8166 236.10608 ]
- [217.16876 348.09952 ]
- [395.43085 142.5729 ]
- ...
- [223.76315 101.57288 ]
- [561.4382 47.980972]
- [200.53334 70.21298 ]]
-05/04/2018 12:36:49 PM - DEBUG - minimum values: [200.00388, 0.00040848268]
-05/04/2018 12:36:49 PM - DEBUG - maximum values: [5745.249, 6567.212]
-05/04/2018 12:36:49 PM - INFO - Creating new RobustScaler
-05/04/2018 12:36:49 PM - INFO - Fitting RobustScaler to training data
-05/04/2018 12:36:49 PM - DEBUG - training data after transformation: [[ 0.7659683 3.400636 ]
- [-0.44083464 5.5718575 ]
- [ 0.9684695 1.5873041 ]
- ...
- [-0.38870072 0.79243493]
- [ 2.28089 -0.24655382]
- [-0.5723509 0.18445913]]
-05/04/2018 12:36:49 PM - INFO - Training and test data transformed
-05/04/2018 12:36:49 PM - INFO - Train model
-05/04/2018 12:36:49 PM - INFO - Using SGD(**{'lr': 0.5, 'momentum': 1.0, 'decay': 0.01}) as Optimizer
-05/04/2018 12:36:49 PM - INFO - Compile model
-05/04/2018 12:36:49 PM - INFO - No weights found, starting completely new model
-05/04/2018 12:36:49 PM - DEBUG - Calculated class_weight: [0.5045154119594469, 55.86593388272504]
-05/04/2018 12:37:35 PM - INFO - Get test loss and metrics of the model
-05/04/2018 12:37:37 PM - INFO - Generation average: 87.82%
-05/04/2018 12:37:37 PM - INFO - --------------------------------------------------------------------------------
-05/04/2018 12:37:37 PM - INFO - ***Doing generation 4 of 7***
-05/04/2018 12:37:37 PM - INFO - Trying to load x_train from ./outputs/x_train.h5
-05/04/2018 12:37:37 PM - INFO - Couldn't load all datasets - reading from ROOT trees
-05/04/2018 12:37:39 PM - DEBUG - training data before transformation: [[369.8166 236.10608 ]
- [217.16876 348.09952 ]
- [395.43085 142.5729 ]
- ...
- [223.76315 101.57288 ]
- [561.4382 47.980972]
- [200.53334 70.21298 ]]
-05/04/2018 12:37:39 PM - DEBUG - minimum values: [200.00388, 0.00040848268]
-05/04/2018 12:37:39 PM - DEBUG - maximum values: [5745.249, 6567.212]
-05/04/2018 12:37:39 PM - INFO - Creating new RobustScaler
-05/04/2018 12:37:39 PM - INFO - Fitting RobustScaler to training data
-05/04/2018 12:37:39 PM - DEBUG - training data after transformation: [[ 0.7659683 3.400636 ]
- [-0.44083464 5.5718575 ]
- [ 0.9684695 1.5873041 ]
- ...
- [-0.38870072 0.79243493]
- [ 2.28089 -0.24655382]
- [-0.5723509 0.18445913]]
-05/04/2018 12:37:39 PM - INFO - Training and test data transformed
-05/04/2018 12:37:39 PM - INFO - Train model
-05/04/2018 12:37:39 PM - INFO - Using SGD(**{'lr': 0.1, 'momentum': 1.0, 'decay': 0.01}) as Optimizer
-05/04/2018 12:37:39 PM - INFO - Compile model
-05/04/2018 12:37:39 PM - INFO - No weights found, starting completely new model
-05/04/2018 12:37:39 PM - DEBUG - Calculated class_weight: [0.5045154119594469, 55.865933882725024]
-05/04/2018 12:40:07 PM - INFO - Get test loss and metrics of the model
-05/04/2018 12:40:11 PM - INFO - Trying to load x_train from ./outputs/x_train.h5
-05/04/2018 12:40:11 PM - INFO - Couldn't load all datasets - reading from ROOT trees
-05/04/2018 12:40:13 PM - DEBUG - training data before transformation: [[369.8166 236.10608 ]
- [217.16876 348.09952 ]
- [395.43085 142.5729 ]
- ...
- [223.76315 101.57288 ]
- [561.4382 47.980972]
- [200.53334 70.21298 ]]
-05/04/2018 12:40:13 PM - DEBUG - minimum values: [200.00388, 0.00040848268]
-05/04/2018 12:40:13 PM - DEBUG - maximum values: [5745.249, 6567.212]
-05/04/2018 12:40:13 PM - INFO - Creating new RobustScaler
-05/04/2018 12:40:13 PM - INFO - Fitting RobustScaler to training data
-05/04/2018 12:40:13 PM - DEBUG - training data after transformation: [[ 0.7659683 3.400636 ]
- [-0.44083464 5.5718575 ]
- [ 0.9684695 1.5873041 ]
- ...
- [-0.38870072 0.79243493]
- [ 2.28089 -0.24655382]
- [-0.5723509 0.18445913]]
-05/04/2018 12:40:13 PM - INFO - Training and test data transformed
-05/04/2018 12:40:13 PM - INFO - Train model
-05/04/2018 12:40:13 PM - INFO - Using SGD(**{'lr': 0.5, 'momentum': 1.0, 'decay': 0.01}) as Optimizer
-05/04/2018 12:40:13 PM - INFO - Compile model
-05/04/2018 12:40:13 PM - INFO - No weights found, starting completely new model
-05/04/2018 12:40:13 PM - DEBUG - Calculated class_weight: [0.5045154119594469, 55.865933882725024]
-05/04/2018 12:42:41 PM - INFO - Get test loss and metrics of the model
-05/04/2018 12:42:45 PM - INFO - Trying to load x_train from ./outputs/x_train.h5
-05/04/2018 12:42:45 PM - INFO - Couldn't load all datasets - reading from ROOT trees
-05/04/2018 12:42:47 PM - DEBUG - training data before transformation: [[369.8166 236.10608 ]
- [217.16876 348.09952 ]
- [395.43085 142.5729 ]
- ...
- [223.76315 101.57288 ]
- [561.4382 47.980972]
- [200.53334 70.21298 ]]
-05/04/2018 12:42:47 PM - DEBUG - minimum values: [200.00388, 0.00040848268]
-05/04/2018 12:42:47 PM - DEBUG - maximum values: [5745.249, 6567.212]
-05/04/2018 12:42:47 PM - INFO - Creating new RobustScaler
-05/04/2018 12:42:47 PM - INFO - Fitting RobustScaler to training data
-05/04/2018 12:42:47 PM - DEBUG - training data after transformation: [[ 0.7659683 3.400636 ]
- [-0.44083464 5.5718575 ]
- [ 0.9684695 1.5873041 ]
- ...
- [-0.38870072 0.79243493]
- [ 2.28089 -0.24655382]
- [-0.5723509 0.18445913]]
-05/04/2018 12:42:47 PM - INFO - Training and test data transformed
-05/04/2018 12:42:47 PM - INFO - Train model
-05/04/2018 12:42:47 PM - INFO - Using SGD(**{'lr': 0.5, 'momentum': 1.0, 'decay': 0.01}) as Optimizer
-05/04/2018 12:42:47 PM - INFO - Compile model
-05/04/2018 12:42:47 PM - INFO - No weights found, starting completely new model
-05/04/2018 12:42:47 PM - DEBUG - Calculated class_weight: [0.5045154119594469, 55.865933882725024]
-05/04/2018 12:45:17 PM - INFO - Get test loss and metrics of the model
-05/04/2018 12:45:21 PM - INFO - Generation average: 97.10%
-05/04/2018 12:45:21 PM - INFO - --------------------------------------------------------------------------------
-05/04/2018 12:45:21 PM - INFO - ***Doing generation 5 of 7***
-05/04/2018 12:45:21 PM - INFO - Trying to load x_train from ./outputs/x_train.h5
-05/04/2018 12:45:21 PM - INFO - Couldn't load all datasets - reading from ROOT trees
-05/04/2018 12:45:23 PM - DEBUG - training data before transformation: [[369.8166 236.10608 ]
- [217.16876 348.09952 ]
- [395.43085 142.5729 ]
- ...
- [223.76315 101.57288 ]
- [561.4382 47.980972]
- [200.53334 70.21298 ]]
-05/04/2018 12:45:23 PM - DEBUG - minimum values: [200.00388, 0.00040848268]
-05/04/2018 12:45:23 PM - DEBUG - maximum values: [5745.249, 6567.212]
-05/04/2018 12:45:23 PM - INFO - Creating new RobustScaler
-05/04/2018 12:45:23 PM - INFO - Fitting RobustScaler to training data
-05/04/2018 12:45:23 PM - DEBUG - training data after transformation: [[ 0.7659683 3.400636 ]
- [-0.44083464 5.5718575 ]
- [ 0.9684695 1.5873041 ]
- ...
- [-0.38870072 0.79243493]
- [ 2.28089 -0.24655382]
- [-0.5723509 0.18445913]]
-05/04/2018 12:45:23 PM - INFO - Training and test data transformed
-05/04/2018 12:45:23 PM - INFO - Train model
-05/04/2018 12:45:23 PM - INFO - Using SGD(**{'lr': 0.1, 'momentum': 1.0, 'decay': 1e-05}) as Optimizer
-05/04/2018 12:45:23 PM - INFO - Compile model
-05/04/2018 12:45:23 PM - INFO - No weights found, starting completely new model
-05/04/2018 12:45:23 PM - DEBUG - Calculated class_weight: [0.5045154119594469, 55.86593388272505]
-05/04/2018 12:47:54 PM - INFO - Get test loss and metrics of the model
-05/04/2018 12:47:59 PM - INFO - Trying to load x_train from ./outputs/x_train.h5
-05/04/2018 12:47:59 PM - INFO - Couldn't load all datasets - reading from ROOT trees
-05/04/2018 12:48:00 PM - DEBUG - training data before transformation: [[369.8166 236.10608 ]
- [217.16876 348.09952 ]
- [395.43085 142.5729 ]
- ...
- [223.76315 101.57288 ]
- [561.4382 47.980972]
- [200.53334 70.21298 ]]
-05/04/2018 12:48:00 PM - DEBUG - minimum values: [200.00388, 0.00040848268]
-05/04/2018 12:48:00 PM - DEBUG - maximum values: [5745.249, 6567.212]
-05/04/2018 12:48:00 PM - INFO - Creating new RobustScaler
-05/04/2018 12:48:00 PM - INFO - Fitting RobustScaler to training data
-05/04/2018 12:48:00 PM - DEBUG - training data after transformation: [[ 0.7659683 3.400636 ]
- [-0.44083464 5.5718575 ]
- [ 0.9684695 1.5873041 ]
- ...
- [-0.38870072 0.79243493]
- [ 2.28089 -0.24655382]
- [-0.5723509 0.18445913]]
-05/04/2018 12:48:00 PM - INFO - Training and test data transformed
-05/04/2018 12:48:00 PM - INFO - Train model
-05/04/2018 12:48:00 PM - INFO - Using SGD(**{'lr': 0.5, 'momentum': 1.0, 'decay': 1e-05}) as Optimizer
-05/04/2018 12:48:00 PM - INFO - Compile model
-05/04/2018 12:48:00 PM - INFO - No weights found, starting completely new model
-05/04/2018 12:48:00 PM - DEBUG - Calculated class_weight: [0.5045154119594469, 55.865933882725024]
-05/04/2018 12:49:05 PM - INFO - Get test loss and metrics of the model
-05/04/2018 12:49:10 PM - INFO - Trying to load x_train from ./outputs/x_train.h5
-05/04/2018 12:49:10 PM - INFO - Couldn't load all datasets - reading from ROOT trees
-05/04/2018 12:49:11 PM - DEBUG - training data before transformation: [[369.8166 236.10608 ]
- [217.16876 348.09952 ]
- [395.43085 142.5729 ]
- ...
- [223.76315 101.57288 ]
- [561.4382 47.980972]
- [200.53334 70.21298 ]]
-05/04/2018 12:49:11 PM - DEBUG - minimum values: [200.00388, 0.00040848268]
-05/04/2018 12:49:11 PM - DEBUG - maximum values: [5745.249, 6567.212]
-05/04/2018 12:49:11 PM - INFO - Creating new RobustScaler
-05/04/2018 12:49:11 PM - INFO - Fitting RobustScaler to training data
-05/04/2018 12:49:11 PM - DEBUG - training data after transformation: [[ 0.7659683 3.400636 ]
- [-0.44083464 5.5718575 ]
- [ 0.9684695 1.5873041 ]
- ...
- [-0.38870072 0.79243493]
- [ 2.28089 -0.24655382]
- [-0.5723509 0.18445913]]
-05/04/2018 12:49:11 PM - INFO - Training and test data transformed
-05/04/2018 12:49:11 PM - INFO - Train model
-05/04/2018 12:49:11 PM - INFO - Using SGD(**{'lr': 0.1, 'momentum': 1.0, 'decay': 1e-05}) as Optimizer
-05/04/2018 12:49:11 PM - INFO - Compile model
-05/04/2018 12:49:11 PM - INFO - No weights found, starting completely new model
-05/04/2018 12:49:11 PM - DEBUG - Calculated class_weight: [0.5045154119594469, 55.86593388272504]
-05/04/2018 12:49:25 PM - INFO - Get test loss and metrics of the model
-05/04/2018 12:49:27 PM - INFO - Generation average: 97.10%
-05/04/2018 12:49:27 PM - INFO - --------------------------------------------------------------------------------
-05/04/2018 12:49:27 PM - INFO - ***Doing generation 6 of 7***
-05/04/2018 12:49:27 PM - INFO - Trying to load x_train from ./outputs/x_train.h5
-05/04/2018 12:49:27 PM - INFO - Couldn't load all datasets - reading from ROOT trees
-05/04/2018 12:49:28 PM - DEBUG - training data before transformation: [[369.8166 236.10608 ]
- [217.16876 348.09952 ]
- [395.43085 142.5729 ]
- ...
- [223.76315 101.57288 ]
- [561.4382 47.980972]
- [200.53334 70.21298 ]]
-05/04/2018 12:49:28 PM - DEBUG - minimum values: [200.00388, 0.00040848268]
-05/04/2018 12:49:28 PM - DEBUG - maximum values: [5745.249, 6567.212]
-05/04/2018 12:49:28 PM - INFO - Creating new RobustScaler
-05/04/2018 12:49:28 PM - INFO - Fitting RobustScaler to training data
-05/04/2018 12:49:28 PM - DEBUG - training data after transformation: [[ 0.7659683 3.400636 ]
- [-0.44083464 5.5718575 ]
- [ 0.9684695 1.5873041 ]
- ...
- [-0.38870072 0.79243493]
- [ 2.28089 -0.24655382]
- [-0.5723509 0.18445913]]
-05/04/2018 12:49:28 PM - INFO - Training and test data transformed
-05/04/2018 12:49:28 PM - INFO - Train model
-05/04/2018 12:49:28 PM - INFO - Using SGD(**{'lr': 0.5, 'momentum': 1.0, 'decay': 1e-05}) as Optimizer
-05/04/2018 12:49:28 PM - INFO - Compile model
-05/04/2018 12:49:28 PM - INFO - No weights found, starting completely new model
-05/04/2018 12:49:28 PM - DEBUG - Calculated class_weight: [0.5045154119594469, 55.865933882725024]
-05/04/2018 12:50:45 PM - INFO - Get test loss and metrics of the model
-05/04/2018 12:50:50 PM - INFO - Generation average: 97.10%
-05/04/2018 12:50:50 PM - INFO - --------------------------------------------------------------------------------
-05/04/2018 12:50:50 PM - INFO - ***Doing generation 7 of 7***
-05/04/2018 12:50:50 PM - INFO - Trying to load x_train from ./outputs/x_train.h5
-05/04/2018 12:50:50 PM - INFO - Couldn't load all datasets - reading from ROOT trees
-05/04/2018 12:50:51 PM - DEBUG - training data before transformation: [[369.8166 236.10608 ]
- [217.16876 348.09952 ]
- [395.43085 142.5729 ]
- ...
- [223.76315 101.57288 ]
- [561.4382 47.980972]
- [200.53334 70.21298 ]]
-05/04/2018 12:50:51 PM - DEBUG - minimum values: [200.00388, 0.00040848268]
-05/04/2018 12:50:51 PM - DEBUG - maximum values: [5745.249, 6567.212]
-05/04/2018 12:50:51 PM - INFO - Creating new RobustScaler
-05/04/2018 12:50:51 PM - INFO - Fitting RobustScaler to training data
-05/04/2018 12:50:51 PM - DEBUG - training data after transformation: [[ 0.7659683 3.400636 ]
- [-0.44083464 5.5718575 ]
- [ 0.9684695 1.5873041 ]
- ...
- [-0.38870072 0.79243493]
- [ 2.28089 -0.24655382]
- [-0.5723509 0.18445913]]
-05/04/2018 12:50:51 PM - INFO - Training and test data transformed
-05/04/2018 12:50:51 PM - INFO - Train model
-05/04/2018 12:50:51 PM - INFO - Using SGD(**{'lr': 1.0, 'momentum': 1.0, 'decay': 0.01}) as Optimizer
-05/04/2018 12:50:51 PM - INFO - Compile model
-05/04/2018 12:50:51 PM - INFO - No weights found, starting completely new model
-05/04/2018 12:50:51 PM - DEBUG - Calculated class_weight: [0.5045154119594469, 55.865933882725024]
-05/04/2018 12:53:28 PM - INFO - Get test loss and metrics of the model
-05/04/2018 12:53:34 PM - INFO - Trying to load x_train from ./outputs/x_train.h5
-05/04/2018 12:53:34 PM - INFO - Couldn't load all datasets - reading from ROOT trees
-05/04/2018 12:53:35 PM - DEBUG - training data before transformation: [[369.8166 236.10608 ]
- [217.16876 348.09952 ]
- [395.43085 142.5729 ]
- ...
- [223.76315 101.57288 ]
- [561.4382 47.980972]
- [200.53334 70.21298 ]]
-05/04/2018 12:53:35 PM - DEBUG - minimum values: [200.00388, 0.00040848268]
-05/04/2018 12:53:35 PM - DEBUG - maximum values: [5745.249, 6567.212]
-05/04/2018 12:53:35 PM - INFO - Creating new RobustScaler
-05/04/2018 12:53:35 PM - INFO - Fitting RobustScaler to training data
-05/04/2018 12:53:35 PM - DEBUG - training data after transformation: [[ 0.7659683 3.400636 ]
- [-0.44083464 5.5718575 ]
- [ 0.9684695 1.5873041 ]
- ...
- [-0.38870072 0.79243493]
- [ 2.28089 -0.24655382]
- [-0.5723509 0.18445913]]
-05/04/2018 12:53:35 PM - INFO - Training and test data transformed
-05/04/2018 12:53:35 PM - INFO - Train model
-05/04/2018 12:53:36 PM - INFO - Using SGD(**{'lr': 0.1, 'momentum': 1.0, 'decay': 1e-05}) as Optimizer
-05/04/2018 12:53:36 PM - INFO - Compile model
-05/04/2018 12:53:36 PM - INFO - No weights found, starting completely new model
-05/04/2018 12:53:36 PM - DEBUG - Calculated class_weight: [0.5045154119594469, 55.865933882725024]
-05/04/2018 12:56:01 PM - INFO - Get test loss and metrics of the model
-05/04/2018 12:56:05 PM - INFO - Generation average: 97.10%
-05/04/2018 12:56:05 PM - INFO - --------------------------------------------------------------------------------
-05/04/2018 12:56:05 PM - INFO - --------------------------------------------------------------------------------
-05/04/2018 12:56:05 PM - INFO - {'decay': 0.01, 'activation': 'relu', 'nb_neurons': 1024, 'lr': 0.1, 'nb_layers': 3, 'momentum': 1.0}
-05/04/2018 12:56:05 PM - INFO - Network accuracy: 97.10%
-05/04/2018 12:56:05 PM - INFO - {'decay': 1e-05, 'activation': 'relu', 'nb_neurons': 1024, 'lr': 0.5, 'nb_layers': 3, 'momentum': 1.0}
-05/04/2018 12:56:05 PM - INFO - Network accuracy: 97.10%
-05/04/2018 12:56:05 PM - INFO - {'decay': 1e-05, 'activation': 'relu', 'nb_neurons': 1024, 'lr': 0.5, 'nb_layers': 3, 'momentum': 1.0}
-05/04/2018 12:56:05 PM - INFO - Network accuracy: 97.10%
-05/04/2018 12:56:05 PM - INFO - {'decay': 0.01, 'activation': 'relu', 'nb_neurons': 1024, 'lr': 1.0, 'nb_layers': 3, 'momentum': 1.0}
-05/04/2018 12:56:05 PM - INFO - Network accuracy: 97.10%
-05/04/2018 12:56:05 PM - INFO - {'decay': 1e-05, 'activation': 'relu', 'nb_neurons': 1024, 'lr': 0.1, 'nb_layers': 3, 'momentum': 1.0}
-05/04/2018 12:56:05 PM - INFO - Network accuracy: 97.10%
diff --git a/Genetic_Algorithm/main.py b/Genetic_Algorithm/main.py
index c1a293b..b73a8f4 100755
--- a/Genetic_Algorithm/main.py
+++ b/Genetic_Algorithm/main.py
@@ -1,119 +1,134 @@
#!/usr/bin/env python
-import logging
-from optimizer import Optimizer
+"""Entry point to evolving the neural network. Start here."""
+from __future__ import print_function
+
+from evolver import Evolver
+
from tqdm import tqdm
+import logging
+
+import sys
+
# Setup logging.
logging.basicConfig(
format='%(asctime)s - %(levelname)s - %(message)s',
datefmt='%m/%d/%Y %I:%M:%S %p',
- level=logging.DEBUG,
+ level=logging.INFO#,
filename='log.txt'
)
-def train_networks(networks):
- """Train each network.
+def train_genomes(genomes):
+ """Train each genome.
Args:
- networks (list): Current population of networks
+ networks (list): Current population of genomes
+
"""
- pbar = tqdm(total=len(networks))
- for network in networks:
- network.train()
+ logging.info("***train_networks(networks)***")
+
+ pbar = tqdm(total=len(genomes))
+
+ for genome in genomes:
+ genome.train()
pbar.update(1)
+
pbar.close()
-def get_average_accuracy(networks):
- """Get the average accuracy for a group of networks.
+def get_average_accuracy(genomes):
+ """Get the average accuracy for a group of networks/genomes.
Args:
- networks (list): List of networks
+ networks (list): List of networks/genomes
Returns:
- float: The average accuracy of a population of networks.
+ float: The average accuracy of a population of networks/genomes.
"""
total_accuracy = 0
- for network in networks:
- total_accuracy += network.accuracy
- return total_accuracy / len(networks)
+ for genome in genomes:
+ total_accuracy += genome.accuracy
-def generate(generations, population, nn_param_choices):
+ return total_accuracy / len(genomes)
+
+def generate(generations, population, all_possible_genes):
"""Generate a network with the genetic algorithm.
Args:
- generations (int): Number of times to evole the population
+ generations (int): Number of times to evolve the population
population (int): Number of networks in each generation
- nn_param_choices (dict): Parameter choices for networks
+ all_possible_genes (dict): Parameter choices for networks
"""
- optimizer = Optimizer(nn_param_choices)
- networks = optimizer.create_population(population)
+ logging.info("***generate(generations, population, all_possible_genes)***")
+
+ evolver = Evolver(all_possible_genes)
+
+ genomes = evolver.create_population(population)
# Evolve the generation.
- for i in range(generations):
- logging.info("***Doing generation %d of %d***" %
- (i + 1, generations))
+ for i in range( generations ):
+
+ logging.info("***Now in generation %d of %d***" % (i + 1, generations))
- # Train and get accuracy for networks.
- train_networks(networks)
+ print_genomes(genomes)
+
+ # Train and get accuracy for networks/genomes.
+ train_genomes(genomes)
# Get the average accuracy for this generation.
- average_accuracy = get_average_accuracy(networks)
+ average_accuracy = get_average_accuracy(genomes)
# Print out the average accuracy each generation.
logging.info("Generation average: %.2f%%" % (average_accuracy * 100))
- logging.info('-'*80)
+ logging.info('-'*80) #-----------
# Evolve, except on the last iteration.
if i != generations - 1:
- # Do the evolution.
- networks = optimizer.evolve(networks)
+ # Evolve!
+ genomes = evolver.evolve(genomes)
- # Sort our final population.
- networks = sorted(networks, key=lambda x: x.accuracy, reverse=True)
+ # Sort our final population according to performance.
+ genomes = sorted(genomes, key=lambda x: x.accuracy, reverse=True)
- # Print out the top 5 networks.
- print_networks(networks[:5])
+ # Print out the top 5 networks/genomes.
+ print_genomes(genomes[:5])
-def print_networks(networks):
- """Print a list of networks.
+ #save_path = saver.save(sess, '/output/model.ckpt')
+ #print("Model saved in file: %s" % save_path)
+
+def print_genomes(genomes):
+ """Print a list of genomes.
Args:
- networks (list): The population of networks
+ genomes (list): The population of networks/genomes
"""
logging.info('-'*80)
- for network in networks:
- network.print_network()
+
+ for genome in genomes:
+ genome.print_genome()
def main():
- """Evolve a network."""
- generations = 7 # Number of times to evolve the population.
- population = 5 # Number of networks in each generation.
-
- nn_param_choices = {
- 'nb_neurons': [8, 16, 32, 64, 128, 256, 512, 768, 1024],
- 'nb_layers': [1, 2, 3, 4, 5, 6, 7, 8, 9, 10],
- 'activation': ['relu', 'elu', 'tanh', 'sigmoid'],
- #'optimizer': ['rmsprop', 'adam', 'sgd', 'adagrad',
- # 'adadelta', 'adamax', 'nadam'],
- #'optimizer_opts': {'lr': [0.1, 0.5, 1.0, 10.0, 100.0],
- # 'decay': [1e-1, 1e-2, 1e-3, 1e-4, 1e-5, 1e-6, 1e-7],
- # 'momentum': [0.0, 0.1, 0.2, 0.3, 0.4, 0.5, 0.6, 0.7,
- # 0.8, 0.9, 1.0]},
- 'lr': [0.1, 0.5, 1.0, 10.0, 100.0],
+ """Evolve a genome."""
+ population = 20 # Number of networks/genomes in each generation.
+ #we only need to train the new ones....
+ generations = 2 # Number of times to evolve the population.
+ all_possible_genes = {
+ 'nb_neurons': [8, 16, 32, 64, 128, 256, 512, 1024],
+ 'nb_layers': [1, 2, 3, 4, 5, 6, 7, 8, 9, 10],
+ 'activation': ['relu', 'elu', 'tanh', 'sigmoid', 'hard_sigmoid','softplus','linear'],
+ #'optimizer': ['rmsprop', 'adam', 'sgd', 'adagrad','adadelta', 'adamax', 'nadam']
+ 'lr': [0.01, 0.05, 0.1, 0.2, 0.3, 0.5, 1.0, 10.0, 100.0],
'decay': [1e-1, 1e-2, 1e-3, 1e-4, 1e-5, 1e-6, 1e-7],
'momentum': [0.0, 0.1, 0.2, 0.3, 0.4, 0.5, 0.6, 0.7, 0.8, 0.9, 1.0]
-
}
- logging.info("***Evolving %d generations with population %d***" %
- (generations, population))
+ print("***Evolving for %d generations with population size = %d***" % (generations, population))
- generate(generations, population, nn_param_choices)
+ generate(generations, population, all_possible_genes)
if __name__ == '__main__':
main()
diff --git a/Genetic_Algorithm/old/main.py b/Genetic_Algorithm/old/main.py
new file mode 100755
index 0000000..c1a293b
--- /dev/null
+++ b/Genetic_Algorithm/old/main.py
@@ -0,0 +1,119 @@
+#!/usr/bin/env python
+
+import logging
+from optimizer import Optimizer
+from tqdm import tqdm
+
+# Setup logging.
+logging.basicConfig(
+ format='%(asctime)s - %(levelname)s - %(message)s',
+ datefmt='%m/%d/%Y %I:%M:%S %p',
+ level=logging.DEBUG,
+ filename='log.txt'
+)
+
+def train_networks(networks):
+ """Train each network.
+
+ Args:
+ networks (list): Current population of networks
+ """
+ pbar = tqdm(total=len(networks))
+ for network in networks:
+ network.train()
+ pbar.update(1)
+ pbar.close()
+
+def get_average_accuracy(networks):
+ """Get the average accuracy for a group of networks.
+
+ Args:
+ networks (list): List of networks
+
+ Returns:
+ float: The average accuracy of a population of networks.
+
+ """
+ total_accuracy = 0
+ for network in networks:
+ total_accuracy += network.accuracy
+
+ return total_accuracy / len(networks)
+
+def generate(generations, population, nn_param_choices):
+ """Generate a network with the genetic algorithm.
+
+ Args:
+ generations (int): Number of times to evole the population
+ population (int): Number of networks in each generation
+ nn_param_choices (dict): Parameter choices for networks
+
+ """
+ optimizer = Optimizer(nn_param_choices)
+ networks = optimizer.create_population(population)
+
+ # Evolve the generation.
+ for i in range(generations):
+ logging.info("***Doing generation %d of %d***" %
+ (i + 1, generations))
+
+ # Train and get accuracy for networks.
+ train_networks(networks)
+
+ # Get the average accuracy for this generation.
+ average_accuracy = get_average_accuracy(networks)
+
+ # Print out the average accuracy each generation.
+ logging.info("Generation average: %.2f%%" % (average_accuracy * 100))
+ logging.info('-'*80)
+
+ # Evolve, except on the last iteration.
+ if i != generations - 1:
+ # Do the evolution.
+ networks = optimizer.evolve(networks)
+
+ # Sort our final population.
+ networks = sorted(networks, key=lambda x: x.accuracy, reverse=True)
+
+ # Print out the top 5 networks.
+ print_networks(networks[:5])
+
+def print_networks(networks):
+ """Print a list of networks.
+
+ Args:
+ networks (list): The population of networks
+
+ """
+ logging.info('-'*80)
+ for network in networks:
+ network.print_network()
+
+def main():
+ """Evolve a network."""
+ generations = 7 # Number of times to evolve the population.
+ population = 5 # Number of networks in each generation.
+
+ nn_param_choices = {
+ 'nb_neurons': [8, 16, 32, 64, 128, 256, 512, 768, 1024],
+ 'nb_layers': [1, 2, 3, 4, 5, 6, 7, 8, 9, 10],
+ 'activation': ['relu', 'elu', 'tanh', 'sigmoid'],
+ #'optimizer': ['rmsprop', 'adam', 'sgd', 'adagrad',
+ # 'adadelta', 'adamax', 'nadam'],
+ #'optimizer_opts': {'lr': [0.1, 0.5, 1.0, 10.0, 100.0],
+ # 'decay': [1e-1, 1e-2, 1e-3, 1e-4, 1e-5, 1e-6, 1e-7],
+ # 'momentum': [0.0, 0.1, 0.2, 0.3, 0.4, 0.5, 0.6, 0.7,
+ # 0.8, 0.9, 1.0]},
+ 'lr': [0.1, 0.5, 1.0, 10.0, 100.0],
+ 'decay': [1e-1, 1e-2, 1e-3, 1e-4, 1e-5, 1e-6, 1e-7],
+ 'momentum': [0.0, 0.1, 0.2, 0.3, 0.4, 0.5, 0.6, 0.7, 0.8, 0.9, 1.0]
+
+ }
+
+ logging.info("***Evolving %d generations with population %d***" %
+ (generations, population))
+
+ generate(generations, population, nn_param_choices)
+
+if __name__ == '__main__':
+ main()
diff --git a/Genetic_Algorithm/network.py b/Genetic_Algorithm/old/network.py
similarity index 100%
rename from Genetic_Algorithm/network.py
rename to Genetic_Algorithm/old/network.py
diff --git a/Genetic_Algorithm/optimizer.py b/Genetic_Algorithm/old/optimizer.py
similarity index 100%
rename from Genetic_Algorithm/optimizer.py
rename to Genetic_Algorithm/old/optimizer.py
diff --git a/Genetic_Algorithm/old/train.py b/Genetic_Algorithm/old/train.py
new file mode 100755
index 0000000..aab4152
--- /dev/null
+++ b/Genetic_Algorithm/old/train.py
@@ -0,0 +1,55 @@
+#!/usr/bin/env python
+
+import os,sys,inspect
+currentdir = os.path.dirname(os.path.abspath(inspect.getfile(inspect.currentframe())))
+parentdir = os.path.dirname(currentdir)
+sys.path.insert(0,parentdir)
+import toolkit
+from toolkit import KerasROOTClassification
+
+def init_model(geneparam):
+
+ nb_layers = geneparam['nb_layers']
+ nb_neurons = geneparam['nb_neurons']
+ activation = geneparam['activation']
+ optimizer = geneparam['optimizer']
+ #lr = network['lr']
+ #decay = network['decay']
+ #momentum = network['momentum']
+
+ filename = "/project/etp4/nhartmann/trees/allTrees_m1.8_NoSys.root"
+
+ c = KerasROOTClassification("",
+ signal_trees = [(filename, "GG_oneStep_1545_1265_985_NoSys")],
+ bkg_trees = [(filename, "ttbar_NoSys"),
+ (filename, "wjets_Sherpa221_NoSys"),
+ (filename, "zjets_Sherpa221_NoSys"),
+ (filename, "diboson_Sherpa221_NoSys"),
+ (filename, "ttv_NoSys"),
+ (filename, "singletop_NoSys")
+ ],
+ dumping_enabled=False,
+ optimizer=optimizer,
+ layers=nb_layers,
+ nodes=nb_neurons,
+ activation_function=activation,
+ # optimizer_opts=dict(lr=lr, decay=decay,
+ # momentum=momentum),
+ earlystopping_opts=dict(monitor='val_loss',
+ min_delta=0, patience=2, verbose=0, mode='auto'),
+ # optimizer="Adam",
+ selection="lep1Pt<5000", # cut out a few very weird outliers
+ branches = ["met", "mt"],
+ weight_expr = "eventWeight*genWeight",
+ identifiers = ["DatasetNumber", "EventNumber"],
+ step_bkg = 100)
+ return c
+
+def train_and_score(geneparam):
+ model = init_model(geneparam)
+
+ model.train(epochs=20)
+
+ score = model.score
+
+ return score[1] # 1 is accuracy. 0 is loss.
diff --git a/Genetic_Algorithm/train.py b/Genetic_Algorithm/train.py
index 4018b05..3e74ba7 100755
--- a/Genetic_Algorithm/train.py
+++ b/Genetic_Algorithm/train.py
@@ -7,15 +7,15 @@ sys.path.insert(0,parentdir)
import toolkit
from toolkit import KerasROOTClassification
-def init_model(network):
+def init_model(geneparam):
- nb_layers = network['nb_layers']
- nb_neurons = network['nb_neurons']
- activation = network['activation']
- # optimizer = network['optimizer']
- lr = network['lr']
- decay = network['decay']
- momentum = network['momentum']
+ nb_layers = geneparam['nb_layers']
+ nb_neurons = geneparam['nb_neurons']
+ activation = geneparam['activation']
+ #optimizer = geneparam['optimizer']
+ lr = geneparam['lr']
+ decay = geneparam['decay']
+ momentum = geneparam['momentum']
filename = "/project/etp4/nhartmann/trees/allTrees_m1.8_NoSys.root"
@@ -28,6 +28,16 @@ def init_model(network):
(filename, "ttv_NoSys"),
(filename, "singletop_NoSys")
],
+ branches = ["jet1Pt", "jet1Phi==-999?0:jet1Phi", "jet1Eta==-999?0:jet1Eta",
+ "jet2Pt", "jet2Phi==-999?0:jet2Phi", "jet2Eta==-999?0:jet2Eta",
+ "jet3Pt", "jet3Phi==-999?0:jet3Phi", "jet3Eta==-999?0:jet3Eta",
+ "jet4Pt", "jet4Phi==-999?0:jet4Phi", "jet4Eta==-999?0:jet4Eta",
+ "jet5Pt", "jet5Phi==-999?0:jet5Phi", "jet5Eta==-999?0:jet5Eta",
+ "jet6Pt", "jet6Phi==-999?0:jet6Phi", "jet6Eta==-999?0:jet6Eta",
+ "jet7Pt", "jet7Phi==-999?0:jet7Phi", "jet7Eta==-999?0:jet7Eta",
+ "jet8Pt", "jet8Phi==-999?0:jet8Phi", "jet8Eta==-999?0:jet8Eta",
+ "lep1Pt", "lep1Phi", "lep1Eta", "nJet30",
+ "met", "met_Phi"],
dumping_enabled=False,
optimizer="SGD",
layers=nb_layers,
@@ -37,19 +47,17 @@ def init_model(network):
momentum=momentum),
earlystopping_opts=dict(monitor='val_loss',
min_delta=0, patience=2, verbose=0, mode='auto'),
- # optimizer="Adam",
selection="lep1Pt<5000", # cut out a few very weird outliers
- branches = ["met", "mt"],
weight_expr = "eventWeight*genWeight",
identifiers = ["DatasetNumber", "EventNumber"],
step_bkg = 100)
return c
-def train_and_score(network):
- model = init_model(network)
+def train_and_score(geneparam):
+ model = init_model(geneparam)
model.train(epochs=20)
-
+ model.evaluate()
score = model.score
return score[1] # 1 is accuracy. 0 is loss.
diff --git a/toolkit.py b/toolkit.py
index bf0a52f..41bff95 100755
--- a/toolkit.py
+++ b/toolkit.py
@@ -17,8 +17,9 @@ from sklearn.preprocessing import StandardScaler, RobustScaler
from sklearn.externals import joblib
from sklearn.metrics import roc_curve, auc
+from keras import backend as K
from keras.models import Sequential
-from keras.layers import Dense
+from keras.layers import Dense, Dropout
from keras.models import model_from_json
from keras.callbacks import History, EarlyStopping
from keras.optimizers import SGD
@@ -343,6 +344,7 @@ class KerasROOTClassification(object):
# the other hidden layers
for layer_number in range(self.layers-1):
self._model.add(Dense(self.nodes, activation=self.activation_function))
+ self._model.add(Dropout(0.2)) # hard-coded dropout for each layer
# last layer is one neuron (binary classification)
self._model.add(Dense(1, activation='sigmoid'))
logger.info("Using {}(**{}) as Optimizer".format(self.optimizer, self.optimizer_opts))
@@ -406,7 +408,7 @@ class KerasROOTClassification(object):
for branch_index, branch in enumerate(self.branches):
self.plot_input(branch_index)
- self.total_epochs = self._read_info("epochs", 0)
+ self.total_epochs = self._read_info("epochs", 0)
logger.info("Train model")
try:
@@ -436,20 +438,28 @@ class KerasROOTClassification(object):
self.total_epochs += epochs
self._write_info("epochs", self.total_epochs)
- #logger.info("Create/Update predictions for ROC curve")
- #self.pred_test = self.model.predict(self.x_test)
- #self.pred_train = self.model.predict(self.x_train)
-
- logger.info("Get test loss and metrics of the model")
- self.score = self.model.evaluate(self.x_test, self.y_test, verbose=0, sample_weight=None)
+ logger.info("Create/Update predictions for ROC curve")
+ self.pred_test = self.model.predict(self.x_test)
+ self.pred_train = self.model.predict(self.x_train)
+
if self.dumping_enabled:
self._dump_to_hdf5("pred_train", "pred_test")
def evaluate(self):
- pass
+ logger.info("Get test loss and metrics of the model")
+ self.score = self.model.evaluate(self.x_test, self.y_test, verbose=0,
+ sample_weight=self.w_train)
+
+ print('Test loss:', self.score[0])
+ print('Test accuracy:', self.score[1])
+
+ #we do not care about keeping any of this in memory -
+ #we just need to know the final scores and the architecture
+ K.clear_session()
+
def write_friend_tree(self):
pass
--
GitLab