experiment.specification package

Submodules

experiment.specification.exp_chapelle module

Created on 27 de mar de 2019

@author: klaus

class experiment.specification.exp_chapelle.ExpChapelle(ds)

Bases: experiment.specification.specification_skeleton.EmptySpecification

classdocs

__init__(ds)

Initialize self. See help(type(self)) for accurate signature.

affmatConfig()

Input configuration.

Returns

List[dict] A list of dictionaries, one for each possible configuration.

algConfig()

Input configuration.

Returns

dict A dictionary, such that each key maps to a list containing each possible value that the attribute might take.

ds = 'g241c'
filterConfig()

Filter configuration.

Returns

List[dict] A list of dictionaries, one for each possible configuration.

generalConfig()

General configuration.

Returns

List[dict] A list of dictionaries, one for each possible configuration.

get_spec_name()

Gets the name identifying the set of experiments that come out of this specification.

inputConfig()

Input configuration.

Returns

List[dict] A list of dictionaries, one for each possible configuration.

noiseConfig()

Noise process configuration.

Returns

List[dict] A list of dictionaries, one for each possible configuration.

class experiment.specification.exp_chapelle.ExpChapelle_2

Bases: experiment.specification.specification_skeleton.EmptySpecification

classdocs

affmatConfig()

Input configuration.

Returns

List[dict] A list of dictionaries, one for each possible configuration.

algConfig()

Input configuration.

Returns

dict A dictionary, such that each key maps to a list containing each possible value that the attribute might take.

filterConfig()

Filter configuration.

Returns

List[dict] A list of dictionaries, one for each possible configuration.

generalConfig()

General configuration.

Returns

List[dict] A list of dictionaries, one for each possible configuration.

get_spec_name()

Gets the name identifying the set of experiments that come out of this specification.

inputConfig()

Input configuration.

Returns

List[dict] A list of dictionaries, one for each possible configuration.

noiseConfig()

Noise process configuration.

Returns

List[dict] A list of dictionaries, one for each possible configuration.

class experiment.specification.exp_chapelle.ExpChapelle_3

Bases: experiment.specification.specification_skeleton.EmptySpecification

classdocs

affmatConfig()

Input configuration.

Returns

List[dict] A list of dictionaries, one for each possible configuration.

algConfig()

Input configuration.

Returns

dict A dictionary, such that each key maps to a list containing each possible value that the attribute might take.

filterConfig()

Filter configuration.

Returns

List[dict] A list of dictionaries, one for each possible configuration.

generalConfig()

General configuration.

Returns

List[dict] A list of dictionaries, one for each possible configuration.

get_spec_name()

Gets the name identifying the set of experiments that come out of this specification.

inputConfig()

Input configuration.

Returns

List[dict] A list of dictionaries, one for each possible configuration.

noiseConfig()

Noise process configuration.

Returns

List[dict] A list of dictionaries, one for each possible configuration.

class experiment.specification.exp_chapelle.ExpChapelle_4

Bases: experiment.specification.specification_skeleton.EmptySpecification

classdocs

affmatConfig()

Input configuration.

Returns

List[dict] A list of dictionaries, one for each possible configuration.

algConfig()

Input configuration.

Returns

dict A dictionary, such that each key maps to a list containing each possible value that the attribute might take.

filterConfig()

Filter configuration.

Returns

List[dict] A list of dictionaries, one for each possible configuration.

generalConfig()

General configuration.

Returns

List[dict] A list of dictionaries, one for each possible configuration.

get_spec_name()

Gets the name identifying the set of experiments that come out of this specification.

inputConfig()

Input configuration.

Returns

List[dict] A list of dictionaries, one for each possible configuration.

noiseConfig()

Noise process configuration.

Returns

List[dict] A list of dictionaries, one for each possible configuration.

class experiment.specification.exp_chapelle.ExpChapelle_5

Bases: experiment.specification.specification_skeleton.EmptySpecification

classdocs

affmatConfig()

Input configuration.

Returns

List[dict] A list of dictionaries, one for each possible configuration.

algConfig()

Input configuration.

Returns

dict A dictionary, such that each key maps to a list containing each possible value that the attribute might take.

filterConfig()

Filter configuration.

Returns

List[dict] A list of dictionaries, one for each possible configuration.

generalConfig()

General configuration.

Returns

List[dict] A list of dictionaries, one for each possible configuration.

get_spec_name()

Gets the name identifying the set of experiments that come out of this specification.

inputConfig()

Input configuration.

Returns

List[dict] A list of dictionaries, one for each possible configuration.

noiseConfig()

Noise process configuration.

Returns

List[dict] A list of dictionaries, one for each possible configuration.

experiment.specification.exp_cifar module

Created on 22 de out de 2019

@author: klaus

class experiment.specification.exp_cifar.ExpCIFAR

Bases: experiment.specification.specification_skeleton.EmptySpecification

classdocs

affmatConfig()

Input configuration.

Returns

List[dict] A list of dictionaries, one for each possible configuration.

algConfig()

Input configuration.

Returns

dict A dictionary, such that each key maps to a list containing each possible value that the attribute might take.

filterConfig()

Filter configuration.

Returns

List[dict] A list of dictionaries, one for each possible configuration.

generalConfig()

General configuration.

Returns

List[dict] A list of dictionaries, one for each possible configuration.

get_spec_name()

Gets the name identifying the set of experiments that come out of this specification.

inputConfig()

Input configuration.

Returns

List[dict] A list of dictionaries, one for each possible configuration.

noiseConfig()

Noise process configuration.

Returns

List[dict] A list of dictionaries, one for each possible configuration.

experiment.specification.exp_debug module

Created on 27 de mar de 2019

@author: klaus

class experiment.specification.exp_debug.ExpDebug

Bases: experiment.specification.specification_skeleton.EmptySpecification

classdocs

affmatConfig()

Input configuration.

Returns

List[dict] A list of dictionaries, one for each possible configuration.

algConfig()

Input configuration.

Returns

dict A dictionary, such that each key maps to a list containing each possible value that the attribute might take.

filterConfig()

Filter configuration.

Returns

List[dict] A list of dictionaries, one for each possible configuration.

generalConfig()

General configuration.

Returns

List[dict] A list of dictionaries, one for each possible configuration.

get_spec_name()

Gets the name identifying the set of experiments that come out of this specification.

inputConfig()

Input configuration.

Returns

List[dict] A list of dictionaries, one for each possible configuration.

noiseConfig()

Noise process configuration.

Returns

List[dict] A list of dictionaries, one for each possible configuration.

experiment.specification.exp_filter_LDST module

Created on 3 de abr de 2019

@author: klaus

class experiment.specification.exp_filter_LDST.FilterLDST

Bases: experiment.specification.specification_skeleton.EmptySpecification

classdocs

WRITE_FREQ = 100
affmatConfig()

Input configuration.

Returns

List[dict] A list of dictionaries, one for each possible configuration.

algConfig()

Input configuration.

Returns

dict A dictionary, such that each key maps to a list containing each possible value that the attribute might take.

filterConfig()

Filter configuration.

Returns

List[dict] A list of dictionaries, one for each possible configuration.

generalConfig()

General configuration.

Returns

List[dict] A list of dictionaries, one for each possible configuration.

get_spec_name()

Gets the name identifying the set of experiments that come out of this specification.

inputConfig()

Input configuration.

Returns

List[dict] A list of dictionaries, one for each possible configuration.

noiseConfig()

Noise process configuration.

Returns

List[dict] A list of dictionaries, one for each possible configuration.

run(cfg)
class experiment.specification.exp_filter_LDST.ISOLET

Bases: experiment.specification.specification_skeleton.EmptySpecification

classdocs

WRITE_FREQ = 100
affmatConfig()

Input configuration.

Returns

List[dict] A list of dictionaries, one for each possible configuration.

algConfig()

Input configuration.

Returns

dict A dictionary, such that each key maps to a list containing each possible value that the attribute might take.

filterConfig()

Filter configuration.

Returns

List[dict] A list of dictionaries, one for each possible configuration.

generalConfig()

General configuration.

Returns

List[dict] A list of dictionaries, one for each possible configuration.

get_spec_name()

Gets the name identifying the set of experiments that come out of this specification.

inputConfig()

Input configuration.

Returns

List[dict] A list of dictionaries, one for each possible configuration.

noiseConfig()

Noise process configuration.

Returns

List[dict] A list of dictionaries, one for each possible configuration.

class experiment.specification.exp_filter_LDST.MNIST

Bases: experiment.specification.specification_skeleton.EmptySpecification

classdocs

WRITE_FREQ = 100
affmatConfig()

Input configuration.

Returns

List[dict] A list of dictionaries, one for each possible configuration.

algConfig()

Input configuration.

Returns

dict A dictionary, such that each key maps to a list containing each possible value that the attribute might take.

filterConfig()

Filter configuration.

Returns

List[dict] A list of dictionaries, one for each possible configuration.

generalConfig()

General configuration.

Returns

List[dict] A list of dictionaries, one for each possible configuration.

get_spec_name()

Gets the name identifying the set of experiments that come out of this specification.

inputConfig()

Input configuration.

Returns

List[dict] A list of dictionaries, one for each possible configuration.

noiseConfig()

Noise process configuration.

Returns

List[dict] A list of dictionaries, one for each possible configuration.

experiment.specification.specification_bits module

Created on 28 de mar de 2019

@author: klaus

experiment.specification.specification_bits.AFFMAT_CIFAR10_LOAD = {'dist_func': ['gaussian'], 'k': [100], 'load_path': ['/home/klaus/eclipse-workspace/NoisyGSSL/src/input/dataset/cifar10_matrices/raw_ANN_k=100.npz'], 'mask_func': ['load'], 'sigma': ['mean']}
experiment.specification.specification_bits.ALGORITHM_NONE = {'algorithm': [None]}
experiment.specification.specification_bits.FILTER_MR = {'filter': ['MRF'], 'p': [1, 4, 15], 'tuning_iter': [1.0, 0.75, 0.5, 0.25, 0], 'tuning_iter_as_pct': [True]}
experiment.specification.specification_bits.FILTER_NOFILTER = {'filter': [None]}
experiment.specification.specification_bits.NOISE_UNIFORM_DET_MODERATE = {'corruption_level': [0.2], 'deterministic': [False], 'type': ['NCAR']}
experiment.specification.specification_bits.add_key_prefix(pref, dictionary)

Adds a prefix to each key in a dictionary.

experiment.specification.specification_bits.allPermutations(args)

Given a dictionary with lists as values, produce every dictionary possible when picking exactly one value from each list.

For example, if there are two keys, each linked to a list of 4 values, the returned list will have size 4*4=16.

Parameters

args (dict) – A dictionary, such that every value is a list.

Returns

A list comprised of every permutation when picking exactly one value from each list of the original dictionary.

Return type

List[dict]

experiment.specification.specification_bits.comb(dict_A, dict_B)

Combine two lists of dictionaries.

Parameters

  • dict_A (List[Dict]) – First list.

  • dict_B (List[Dict]) – Second list.

Returns

all possible dictionaries when extending a dict from the 1st list with one from the 2nd.

Return type

List[Dict]

experiment.specification.specification_skeleton module

Created on 27 de mar de 2019

@author: klaus

class experiment.specification.specification_skeleton.EmptySpecification

Bases: object

EmptySpecification defines the methods expected from any class representing a specification of experiments. By itself, it also specifies an empty set of experiments.

DEBUG_MODE = True
FORCE_GTAM_LDST_SAME_MU = True
OVERWRITE = True
TUNING_ITER_AS_NOISE_PCT = False
WRITE_FREQ = 10000
affmatConfig()

Input configuration.

Returns

List[dict] A list of dictionaries, one for each possible configuration.

aggregate_csv()
algConfig()

Input configuration.

Returns

dict A dictionary, such that each key maps to a list containing each possible value that the attribute might take.

filterConfig()

Filter configuration.

Returns

List[dict] A list of dictionaries, one for each possible configuration.

generalConfig()

General configuration.

Returns

List[dict] A list of dictionaries, one for each possible configuration.

get_all_configs()

Gets the configuration for every experiment. The corresponding prefix is added for each stage. Returns: List[dict] A list of all possible configs.

get_spec_name()

Gets the name identifying the set of experiments that come out of this specification.

inputConfig()

Input configuration.

Returns

List[dict] A list of dictionaries, one for each possible configuration.

noiseConfig()

Noise process configuration.

Returns

List[dict] A list of dictionaries, one for each possible configuration.

run(cfg)
run_all()
experiment.specification.specification_skeleton.processify(func)

From https://gist.github.com/schlamar/2311116 Decorator to run a function as a process. Be sure that every argument and the return value is pickable. The created process is joined, so the code does not run in parallel.

experiment.specification.specification_skeleton.runprocessify_func(q, *args, **kwargs)

Module contents