Compare commits

...

11 Commits

Author SHA1 Message Date
Alexander Engelsberger
391473adf3 build: bump version 0.7.5 → 0.7.6 2023-10-04 14:47:27 +02:00
Alexander Engelsberger
0d8db31ff2
ci: update python versions 2023-06-20 16:34:41 +02:00
Alexander Engelsberger
89b96f0a98
chore: switch to pytorch 2.0+ 2023-06-20 16:27:54 +02:00
Alexander Engelsberger
ee4cf583e3
chore: fix minor errors and upgrade codebase 2023-06-20 16:06:53 +02:00
Alexander Engelsberger
6ed1b9a832
feat: add gmlvq example
it was necessary to update the pre-commit definition for a successfull
commit.
2023-06-20 15:12:32 +02:00
Alexander Engelsberger
4a7d4a3d99
chore(ci): update github actions 2022-12-05 17:14:54 +01:00
Alexander Engelsberger
0626af207f
build: bump version 0.7.4 → 0.7.5 2022-12-05 17:03:04 +01:00
rmschubert
7b23983887 fix: update scikit-learn dependency 2022-12-05 16:48:22 +01:00
Alexander Engelsberger
0649d5bb45
build: bump version 0.7.3 → 0.7.4 2022-05-17 11:57:32 +02:00
Alexander Engelsberger
339316aa7e
fix: use epsilon in cbc competition 2022-05-17 11:56:43 +02:00
Alexander Engelsberger
2a85c94b55
chore: minor changes and version updates 2022-05-17 11:56:18 +02:00
16 changed files with 230 additions and 136 deletions

View File

@ -1,5 +1,5 @@
[bumpversion]
current_version = 0.7.3
current_version = 0.7.6
commit = True
tag = True
parse = (?P<major>\d+)\.(?P<minor>\d+)\.(?P<patch>\d+)

View File

@ -6,70 +6,70 @@ name: tests
on:
push:
pull_request:
branches: [ master ]
branches: [master]
jobs:
style:
runs-on: ubuntu-latest
steps:
- uses: actions/checkout@v2
- name: Set up Python 3.10
uses: actions/setup-python@v2
with:
python-version: "3.10"
- name: Install dependencies
run: |
python -m pip install --upgrade pip
pip install .[all]
- uses: pre-commit/action@v2.0.3
- uses: actions/checkout@v3
- name: Set up Python 3.11
uses: actions/setup-python@v4
with:
python-version: "3.11"
- name: Install dependencies
run: |
python -m pip install --upgrade pip
pip install .[all]
- uses: pre-commit/action@v3.0.0
compatibility:
needs: style
strategy:
fail-fast: false
matrix:
python-version: ["3.7", "3.8", "3.9", "3.10"]
python-version: ["3.8", "3.9", "3.10", "3.11"]
os: [ubuntu-latest, windows-latest]
exclude:
- os: windows-latest
python-version: "3.7"
- os: windows-latest
python-version: "3.8"
- os: windows-latest
python-version: "3.9"
- os: windows-latest
python-version: "3.8"
- os: windows-latest
python-version: "3.9"
- os: windows-latest
python-version: "3.10"
runs-on: ${{ matrix.os }}
steps:
- uses: actions/checkout@v2
- name: Set up Python ${{ matrix.python-version }}
uses: actions/setup-python@v2
with:
python-version: ${{ matrix.python-version }}
- name: Install dependencies
run: |
python -m pip install --upgrade pip
pip install .[all]
- name: Test with pytest
run: |
pytest
- uses: actions/checkout@v3
- name: Set up Python ${{ matrix.python-version }}
uses: actions/setup-python@v4
with:
python-version: ${{ matrix.python-version }}
- name: Install dependencies
run: |
python -m pip install --upgrade pip
pip install .[all]
- name: Test with pytest
run: |
pytest
publish_pypi:
if: github.event_name == 'push' && startsWith(github.ref, 'refs/tags')
needs: compatibility
runs-on: ubuntu-latest
steps:
- uses: actions/checkout@v2
- name: Set up Python 3.10
uses: actions/setup-python@v2
with:
python-version: "3.10"
- name: Install dependencies
run: |
python -m pip install --upgrade pip
pip install .[all]
pip install wheel
- name: Build package
run: python setup.py sdist bdist_wheel
- name: Publish a Python distribution to PyPI
uses: pypa/gh-action-pypi-publish@release/v1
with:
user: __token__
password: ${{ secrets.PYPI_API_TOKEN }}
- uses: actions/checkout@v3
- name: Set up Python 3.10
uses: actions/setup-python@v4
with:
python-version: "3.11"
- name: Install dependencies
run: |
python -m pip install --upgrade pip
pip install .[all]
pip install wheel
- name: Build package
run: python setup.py sdist bdist_wheel
- name: Publish a Python distribution to PyPI
uses: pypa/gh-action-pypi-publish@release/v1
with:
user: __token__
password: ${{ secrets.PYPI_API_TOKEN }}

View File

@ -3,7 +3,7 @@
repos:
- repo: https://github.com/pre-commit/pre-commit-hooks
rev: v4.1.0
rev: v4.4.0
hooks:
- id: trailing-whitespace
- id: end-of-file-fixer
@ -13,17 +13,17 @@ repos:
- id: check-case-conflict
- repo: https://github.com/myint/autoflake
rev: v1.4
rev: v2.1.1
hooks:
- id: autoflake
- repo: http://github.com/PyCQA/isort
rev: 5.10.1
rev: 5.12.0
hooks:
- id: isort
- repo: https://github.com/pre-commit/mirrors-mypy
rev: v0.931
rev: v1.3.0
hooks:
- id: mypy
files: prototorch
@ -35,14 +35,14 @@ repos:
- id: yapf
- repo: https://github.com/pre-commit/pygrep-hooks
rev: v1.9.0
rev: v1.10.0
hooks:
- id: python-use-type-annotations
- id: python-no-log-warn
- id: python-check-blanket-noqa
- repo: https://github.com/asottile/pyupgrade
rev: v2.31.0
rev: v3.7.0
hooks:
- id: pyupgrade

View File

@ -23,7 +23,7 @@ author = "Jensun Ravichandran"
# The full version, including alpha/beta/rc tags
#
release = "0.7.3"
release = "0.7.6"
# -- General configuration ---------------------------------------------------

View File

@ -1,5 +1,7 @@
"""ProtoTorch CBC example using 2D Iris data."""
import logging
import torch
from matplotlib import pyplot as plt
@ -34,7 +36,7 @@ class VisCBC2D():
self.resolution = 100
self.cmap = "viridis"
def on_epoch_end(self):
def on_train_epoch_end(self):
x_train, y_train = self.x_train, self.y_train
_components = self.model.components_layer._components.detach()
ax = self.fig.gca()
@ -94,5 +96,5 @@ if __name__ == "__main__":
correct += (y_pred.argmax(1) == y).float().sum(0)
acc = 100 * correct / len(train_ds)
print(f"Epoch: {epoch} Accuracy: {acc:05.02f}%")
vis.on_epoch_end()
logging.info(f"Epoch: {epoch} Accuracy: {acc:05.02f}%")
vis.on_train_epoch_end()

76
examples/gmlvq.py Normal file
View File

@ -0,0 +1,76 @@
"""ProtoTorch GMLVQ example using Iris data."""
import torch
import prototorch as pt
class GMLVQ(torch.nn.Module):
"""
Implementation of Generalized Matrix Learning Vector Quantization.
"""
def __init__(self, data, **kwargs):
super().__init__(**kwargs)
self.components_layer = pt.components.LabeledComponents(
distribution=[1, 1, 1],
components_initializer=pt.initializers.SMCI(data, noise=0.1),
)
self.backbone = pt.transforms.Omega(
len(data[0][0]),
len(data[0][0]),
pt.initializers.RandomLinearTransformInitializer(),
)
def forward(self, data):
"""
Forward function that returns a tuple of dissimilarities and label information.
Feed into GLVQLoss to get a complete GMLVQ model.
"""
components, label = self.components_layer()
latent_x = self.backbone(data)
latent_components = self.backbone(components)
distance = pt.distances.squared_euclidean_distance(
latent_x, latent_components)
return distance, label
def predict(self, data):
"""
The GMLVQ has a modified prediction step, where a competition layer is applied.
"""
components, label = self.components_layer()
distance = pt.distances.squared_euclidean_distance(data, components)
winning_label = pt.competitions.wtac(distance, label)
return winning_label
if __name__ == "__main__":
train_ds = pt.datasets.Iris()
train_loader = torch.utils.data.DataLoader(train_ds, batch_size=32)
model = GMLVQ(train_ds)
optimizer = torch.optim.Adam(model.parameters(), lr=0.05)
criterion = pt.losses.GLVQLoss()
for epoch in range(200):
correct = 0.0
for x, y in train_loader:
d, labels = model(x)
loss = criterion(d, y, labels).mean(0)
optimizer.zero_grad()
loss.backward()
optimizer.step()
with torch.no_grad():
y_pred = model.predict(x)
correct += (y_pred == y).float().sum(0)
acc = 100 * correct / len(train_ds)
print(f"Epoch: {epoch} Accuracy: {acc:05.02f}%")

View File

@ -17,7 +17,7 @@ from .core import similarities # noqa: F401
from .core import transforms # noqa: F401
# Core Setup
__version__ = "0.7.3"
__version__ = "0.7.6"
__all_core__ = [
"competitions",

View File

@ -38,7 +38,7 @@ def cbcc(detections: torch.Tensor, reasonings: torch.Tensor):
pk = A
nk = (1 - A) * B
numerator = (detections @ (pk - nk).T) + nk.sum(1)
probs = numerator / (pk + nk).sum(1)
probs = numerator / ((pk + nk).sum(1) + 1e-8)
return probs

View File

@ -11,7 +11,7 @@ def squared_euclidean_distance(x, y):
**Alias:**
``prototorch.functions.distances.sed``
"""
x, y = [arr.view(arr.size(0), -1) for arr in (x, y)]
x, y = (arr.view(arr.size(0), -1) for arr in (x, y))
expanded_x = x.unsqueeze(dim=1)
batchwise_difference = y - expanded_x
differences_raised = torch.pow(batchwise_difference, 2)
@ -27,14 +27,14 @@ def euclidean_distance(x, y):
:returns: Distance Tensor of shape :math:`X \times Y`
:rtype: `torch.tensor`
"""
x, y = [arr.view(arr.size(0), -1) for arr in (x, y)]
x, y = (arr.view(arr.size(0), -1) for arr in (x, y))
distances_raised = squared_euclidean_distance(x, y)
distances = torch.sqrt(distances_raised)
return distances
def euclidean_distance_v2(x, y):
x, y = [arr.view(arr.size(0), -1) for arr in (x, y)]
x, y = (arr.view(arr.size(0), -1) for arr in (x, y))
diff = y - x.unsqueeze(1)
pairwise_distances = (diff @ diff.permute((0, 2, 1))).sqrt()
# Passing `dim1=-2` and `dim2=-1` to `diagonal()` takes the
@ -54,7 +54,7 @@ def lpnorm_distance(x, y, p):
:param p: p parameter of the lp norm
"""
x, y = [arr.view(arr.size(0), -1) for arr in (x, y)]
x, y = (arr.view(arr.size(0), -1) for arr in (x, y))
distances = torch.cdist(x, y, p=p)
return distances
@ -66,7 +66,7 @@ def omega_distance(x, y, omega):
:param `torch.tensor` omega: Two dimensional matrix
"""
x, y = [arr.view(arr.size(0), -1) for arr in (x, y)]
x, y = (arr.view(arr.size(0), -1) for arr in (x, y))
projected_x = x @ omega
projected_y = y @ omega
distances = squared_euclidean_distance(projected_x, projected_y)
@ -80,7 +80,7 @@ def lomega_distance(x, y, omegas):
:param `torch.tensor` omegas: Three dimensional matrix
"""
x, y = [arr.view(arr.size(0), -1) for arr in (x, y)]
x, y = (arr.view(arr.size(0), -1) for arr in (x, y))
projected_x = x @ omegas
projected_y = torch.diagonal(y @ omegas).T
expanded_y = torch.unsqueeze(projected_y, dim=1)

View File

@ -21,7 +21,7 @@ def cosine_similarity(x, y):
Expected dimension of x is 2.
Expected dimension of y is 2.
"""
x, y = [arr.view(arr.size(0), -1) for arr in (x, y)]
x, y = (arr.view(arr.size(0), -1) for arr in (x, y))
norm_x = x.pow(2).sum(1).sqrt()
norm_y = y.pow(2).sum(1).sqrt()
norm_mat = norm_x.unsqueeze(-1) @ norm_y.unsqueeze(-1).T

View File

@ -20,7 +20,7 @@ class Dataset(torch.utils.data.Dataset):
_repr_indent = 2
def __init__(self, root):
if isinstance(root, torch._six.string_classes):
if isinstance(root, str):
root = os.path.expanduser(root)
self.root = root

View File

@ -5,8 +5,10 @@ URL:
"""
from __future__ import annotations
import warnings
from typing import Sequence, Union
from typing import Sequence
from sklearn.datasets import (
load_iris,
@ -41,9 +43,9 @@ class Iris(NumpyDataset):
:param dims: select a subset of dimensions
"""
def __init__(self, dims: Sequence[int] = None):
def __init__(self, dims: Sequence[int] | None = None):
x, y = load_iris(return_X_y=True)
if dims:
if dims is not None:
x = x[:, dims]
super().__init__(x, y)
@ -56,15 +58,19 @@ class Blobs(NumpyDataset):
"""
def __init__(self,
num_samples: int = 300,
num_features: int = 2,
seed: Union[None, int] = 0):
x, y = make_blobs(num_samples,
num_features,
centers=None,
random_state=seed,
shuffle=False)
def __init__(
self,
num_samples: int = 300,
num_features: int = 2,
seed: None | int = 0,
):
x, y = make_blobs(
num_samples,
num_features,
centers=None,
random_state=seed,
shuffle=False,
)
super().__init__(x, y)
@ -77,29 +83,33 @@ class Random(NumpyDataset):
Note: n_classes * n_clusters_per_class <= 2**n_informative must satisfy.
"""
def __init__(self,
num_samples: int = 300,
num_features: int = 2,
num_classes: int = 2,
num_clusters: int = 2,
num_informative: Union[None, int] = None,
separation: float = 1.0,
seed: Union[None, int] = 0):
def __init__(
self,
num_samples: int = 300,
num_features: int = 2,
num_classes: int = 2,
num_clusters: int = 2,
num_informative: None | int = None,
separation: float = 1.0,
seed: None | int = 0,
):
if not num_informative:
import math
num_informative = math.ceil(math.log2(num_classes * num_clusters))
if num_features < num_informative:
warnings.warn("Generating more features than requested.")
num_features = num_informative
x, y = make_classification(num_samples,
num_features,
n_informative=num_informative,
n_redundant=0,
n_classes=num_classes,
n_clusters_per_class=num_clusters,
class_sep=separation,
random_state=seed,
shuffle=False)
x, y = make_classification(
num_samples,
num_features,
n_informative=num_informative,
n_redundant=0,
n_classes=num_classes,
n_clusters_per_class=num_clusters,
class_sep=separation,
random_state=seed,
shuffle=False,
)
super().__init__(x, y)
@ -113,16 +123,20 @@ class Circles(NumpyDataset):
"""
def __init__(self,
num_samples: int = 300,
noise: float = 0.3,
factor: float = 0.8,
seed: Union[None, int] = 0):
x, y = make_circles(num_samples,
noise=noise,
factor=factor,
random_state=seed,
shuffle=False)
def __init__(
self,
num_samples: int = 300,
noise: float = 0.3,
factor: float = 0.8,
seed: None | int = 0,
):
x, y = make_circles(
num_samples,
noise=noise,
factor=factor,
random_state=seed,
shuffle=False,
)
super().__init__(x, y)
@ -136,12 +150,16 @@ class Moons(NumpyDataset):
"""
def __init__(self,
num_samples: int = 300,
noise: float = 0.3,
seed: Union[None, int] = 0):
x, y = make_moons(num_samples,
noise=noise,
random_state=seed,
shuffle=False)
def __init__(
self,
num_samples: int = 300,
noise: float = 0.3,
seed: None | int = 0,
):
x, y = make_moons(
num_samples,
noise=noise,
random_state=seed,
shuffle=False,
)
super().__init__(x, y)

View File

@ -36,6 +36,7 @@ Description:
are determined by analytic chemistry.
"""
import logging
import os
import numpy as np
@ -81,13 +82,11 @@ class Tecator(ProtoDataset):
if self._check_exists():
return
if self.verbose:
print("Making directories...")
logging.debug("Making directories...")
os.makedirs(self.raw_folder, exist_ok=True)
os.makedirs(self.processed_folder, exist_ok=True)
if self.verbose:
print("Downloading...")
logging.debug("Downloading...")
for fileid, md5 in self._resources:
filename = "tecator.npz"
download_file_from_google_drive(fileid,
@ -95,8 +94,7 @@ class Tecator(ProtoDataset):
filename=filename,
md5=md5)
if self.verbose:
print("Processing...")
logging.debug("Processing...")
with np.load(os.path.join(self.raw_folder, "tecator.npz"),
allow_pickle=False) as f:
x_train, y_train = f["x_train"], f["y_train"]
@ -117,5 +115,4 @@ class Tecator(ProtoDataset):
"wb") as f:
torch.save(test_set, f)
if self.verbose:
print("Done!")
logging.debug("Done!")

View File

@ -5,6 +5,7 @@ from typing import (
Dict,
Iterable,
List,
Optional,
Union,
)
@ -18,7 +19,7 @@ def generate_mesh(
maxima: torch.TensorType,
border: float = 1.0,
resolution: int = 100,
device: torch.device = None,
device: Optional[torch.device] = None,
):
# Apply Border
ptp = maxima - minima
@ -55,14 +56,15 @@ def mesh2d(x=None, border: float = 1.0, resolution: int = 100):
def distribution_from_list(list_dist: List[int],
clabels: Iterable[int] = None):
clabels: Optional[Iterable[int]] = None):
clabels = clabels or list(range(len(list_dist)))
distribution = dict(zip(clabels, list_dist))
return distribution
def parse_distribution(user_distribution,
clabels: Iterable[int] = None) -> Dict[int, int]:
def parse_distribution(
user_distribution,
clabels: Optional[Iterable[int]] = None) -> Dict[int, int]:
"""Parse user-provided distribution.
Return a dictionary with integer keys that represent the class labels and

View File

@ -15,14 +15,14 @@ from setuptools import find_packages, setup
PROJECT_URL = "https://github.com/si-cim/prototorch"
DOWNLOAD_URL = "https://github.com/si-cim/prototorch.git"
with open("README.md", "r") as fh:
with open("README.md", encoding="utf-8") as fh:
long_description = fh.read()
INSTALL_REQUIRES = [
"torch>=1.3.1",
"torchvision>=0.7.3",
"numpy>=1.9.1",
"sklearn",
"torch>=2.0.0",
"torchvision",
"numpy",
"scikit-learn",
"matplotlib",
]
DATASETS = [
@ -51,7 +51,7 @@ ALL = DATASETS + DEV + DOCS + EXAMPLES + TESTS
setup(
name="prototorch",
version="0.7.3",
version="0.7.6",
description="Highly extensible, GPU-supported "
"Learning Vector Quantization (LVQ) toolbox "
"built using PyTorch and its nn API.",
@ -62,7 +62,7 @@ setup(
url=PROJECT_URL,
download_url=DOWNLOAD_URL,
license="MIT",
python_requires=">=3.7",
python_requires=">=3.8",
install_requires=INSTALL_REQUIRES,
extras_require={
"datasets": DATASETS,
@ -85,10 +85,10 @@ setup(
"License :: OSI Approved :: MIT License",
"Operating System :: OS Independent",
"Programming Language :: Python :: 3",
"Programming Language :: Python :: 3.7",
"Programming Language :: Python :: 3.8",
"Programming Language :: Python :: 3.9",
"Programming Language :: Python :: 3.10",
"Programming Language :: Python :: 3.11",
],
packages=find_packages(),
zip_safe=False,

View File

@ -1,7 +1,6 @@
"""ProtoTorch datasets test suite"""
import os
import shutil
import unittest
import numpy as np