Commit a65fffe6 authored by Rémi Huguet's avatar Rémi Huguet
Browse files

feat: Add MIA distinguisher and analysis APIs.

parent 711ff5b8

.gitlab-ci.yml

+1 −1
Original line number Diff line number Diff line
@@ -11,7 +11,7 @@ lint: 
  script:

    - conda install python=3.6

    - pip install -e .

    - pip install flake8 pep8-naming flake8-docstrings

    - pip install "pydocstyle<4" flake8 pep8-naming flake8-docstrings

    - flake8 --config=.flake8 scared tests setup.py



test:

docs/source/api_reference/distinguishers/mia.rst

0 → 100644
+8 −0
Original line number Diff line number Diff line 
Distinguishers MIA API reference

================================



.. toctree::

   :maxdepth: 1



.. automodule:: scared.distinguishers.mia

    :members:

scared/__init__.py

+7 −2
Original line number Diff line number Diff line
@@ -13,10 +13,15 @@ from .distinguishers import ( # noqa: F401 
    DPADistinguisherMixin, DPADistinguisher,

    CPADistinguisherMixin, CPAAlternativeDistinguisherMixin, CPADistinguisher, CPAAlternativeDistinguisher,

    PartitionedDistinguisherMixin, PartitionedDistinguisher, ANOVADistinguisherMixin, ANOVADistinguisher,

    NICVDistinguisherMixin, NICVDistinguisher, SNRDistinguisherMixin, SNRDistinguisher

    NICVDistinguisherMixin, NICVDistinguisher, SNRDistinguisherMixin, SNRDistinguisher,

    MIADistinguisher

)

from .ttest import TTestAccumulator, TTestAnalysis, TTestError, TTestContainer  # noqa:F401

from .analysis import BaseAnalysis, CPAAnalysis, DPAAnalysis, ANOVAAnalysis, NICVAnalysis, SNRAnalysis, BasePartitionedAnalysis  # noqa:F401

from .analysis import (  # noqa:F401

    BaseAnalysis, CPAAnalysis, DPAAnalysis,

    ANOVAAnalysis, NICVAnalysis, SNRAnalysis,

    BasePartitionedAnalysis, MIAAnalysis

)

from .preprocesses import preprocess, Preprocess, PreprocessError  # noqa:F401

from . import container as _container

scared/analysis.py

+26 −2
Original line number Diff line number Diff line 
from . import selection_functions as _sf, container as _container, models, distinguishers

import numpy as _np

import logging



logger = logging.getLogger(__name__)





class BaseAnalysis:
@@ -98,6 +101,7 @@ class BaseAnalysis: 
        return base_batch_size



    def _final_compute(self):

        logger.info(f'Starting final computing.')

        self.compute_results()

        if self.convergence_step and len(self._batches_processed) > 1:

            self._compute_convergence_traces()
@@ -118,10 +122,14 @@ class BaseAnalysis: 
            raise TypeError(f'container should be a type Container, not {type(container)}.')



        batch_size = self._compute_batch_size(container.batch_size)

        for batch in container.batches(batch_size=batch_size):

        logger.info(f'Starting run on container {container}, with batch size {batch_size}.')

        for i, batch in enumerate(container.batches(batch_size=batch_size)):

            logger.info(f'Process batch number {i} starting.')

            self.process(batch)

            self.compute_convergence()

        logger.info(f'Batches processing finished.')

        self._final_compute()

        logger.info(f'Run on container {container} finished.')



    def compute_intermediate_values(self, metadata):

        """Compute intermediate leakage values for this instance from metadata.
@@ -130,6 +138,7 @@ class BaseAnalysis: 
            metadata (mapping): a dict-like object containing the data to be used with selection function.



        """

        logger.info(f'Computing intermediate values for metadata {metadata}.')

        return self.model(self.selection_function(**metadata))



    def process(self, traces_batch):
@@ -146,8 +155,11 @@ class BaseAnalysis: 
        intermediate_values = self.compute_intermediate_values(traces_batch.metadatas)



        if self.convergence_traces is None and self.convergence_step:

            logger.info(f'Initialize convergence traces.')

            self.convergence_traces = _np.empty(intermediate_values.shape[1:] + (0, ), dtype=self.precision)



        logger.info(f'Will call distinguisher update with {traces_batch}.')



        self.update(

            data=intermediate_values,

            traces=traces_batch.samples
@@ -159,6 +171,7 @@ class BaseAnalysis: 
        This method is used internally by `run`, but can also be used to have a finer control on the process.



        """

        logger.info(f'Compute convergence results.')

        if self.convergence_step:

            self._batches_processed.append(self.processed_traces)

            if self._batches_processed[-1] - self._batches_processed[0] >= self.convergence_step:
@@ -167,6 +180,7 @@ class BaseAnalysis: 
                self._compute_convergence_traces()



    def _compute_convergence_traces(self):

        logger.info(f'Update convergence traces.')

        self.convergence_traces = _np.append(self.convergence_traces, self.scores[..., None], axis=-1)



    def compute_results(self):
@@ -175,8 +189,10 @@ class BaseAnalysis: 
        This method is used internally by `run`, but can also be used to have a finer control on the process.



        """

        logger.info(f'Computing results ...')

        self.results = self.compute()

        self.scores = self.discriminant(self.results)

        logger.info(f'Results computed.')





class CPAAnalysis(BaseAnalysis, distinguishers.CPADistinguisherMixin):
@@ -195,8 +211,8 @@ class DPAAnalysis(BaseAnalysis, distinguishers.DPADistinguisherMixin): 


class BasePartitionedAnalysis(BaseAnalysis):

    def __init__(self, partitions=None, *args, **kwargs):

        super().__init__(*args, **kwargs)

        distinguishers.partitioned._set_partitions(self, partitions)

        return super().__init__(*args, **kwargs)





class ANOVAAnalysis(BasePartitionedAnalysis, distinguishers.ANOVADistinguisherMixin):
@@ -209,3 +225,11 @@ class NICVAnalysis(BasePartitionedAnalysis, distinguishers.NICVDistinguisherMixi 


class SNRAnalysis(BasePartitionedAnalysis, distinguishers.SNRDistinguisherMixin):

    __doc__ = distinguishers.SNRDistinguisherMixin.__doc__ + BaseAnalysis.__doc__





class MIAAnalysis(BasePartitionedAnalysis, distinguishers.MIADistinguisherMixin):

    __doc__ = distinguishers.MIADistinguisherMixin.__doc__ + BaseAnalysis.__doc__



    def __init__(self, bins_number=128, bin_edges=None, *args, **kwargs):

        distinguishers.mia._set_histogram_parameters(self, bins_number=bins_number, bin_edges=bin_edges)

        return super().__init__(*args, **kwargs)

scared/distinguishers/__init__.py

+1 −0
Original line number Diff line number Diff line
@@ -7,3 +7,4 @@ from .partitioned import ( # noqa: F401 
    SNRDistinguisher, SNRDistinguisherMixin,

    NICVDistinguisher, NICVDistinguisherMixin

)

from .mia import MIADistinguisher, MIADistinguisherMixin  # noqa: F401