lgeiger

import numpy as np
import matplotlib.pyplot as plt
from plotly import offline as py
import plotly.tools as tls
py.init_notebook_mode()

t = np.linspace(0, 10, 1000)
plt.plot(t, np.exp(-0.5 * t) * np.cos(2*np.pi*t))
plt.xlim(0, 7)

lgeiger

import tensorflow as tf
from groupy.gconv.gconv_tensorflow.keras.layers import P4ConvZ2, P4ConvP4

batch_size = 32
num_classes = 10
epochs = 25
num_predictions = 20

# The data, split between train and test sets:
(x_train, y_train), (x_test, y_test) = tf.keras.datasets.cifar10.load_data()

lgeiger

Hyperparameter optimization using law and scikit optimize

lgeiger

commit 1ebd8b170d31d64ad2523a1db81c5619fed24fc1
Author: Lukas Geiger <[email protected]>
Date:   Sat Mar 31 01:40:36 2018 +0200

    one_sided penalty

diff --git a/tensorflow/contrib/gan/python/losses/python/losses_impl.py b/tensorflow/contrib/gan/python/losses/python/losses_impl.py
index 2a40dbade6..77a86043d8 100644
--- a/tensorflow/contrib/gan/python/losses/python/losses_impl.py
+++ b/tensorflow/contrib/gan/python/losses/python/losses_impl.py

lgeiger

from __future__ import absolute_import
from __future__ import division
from __future__ import print_function

from tensorflow.python.eager import context
from tensorflow.python.ops import control_flow_ops
from tensorflow.python.ops import math_ops
from tensorflow.python.ops import state_ops
from tensorflow.python.framework import ops
from tensorflow.python.training import optimizer

lgeiger

{
  "citation": "@inproceedings{Silberman:ECCV12,\n  author    = {Nathan Silberman, Derek Hoiem, Pushmeet Kohli and Rob Fergus},\n  title     = {Indoor Segmentation and Support Inference from RGBD Images},\n  booktitle = {ECCV},\n  year      = {2012}\n}\n@article{Alhashim2018,\n  author    = {Ibraheem Alhashim and Peter Wonka},\n  title     = {High Quality Monocular Depth Estimation via Transfer Learning},\n  journal   = {arXiv e-prints},\n  volume    = {abs/1812.11941},\n  year      = {2018},\n  url       = {https://arxiv.org/abs/1812.11941},\n  eid       = {arXiv:1812.11941},\n  eprint    = {1812.11941}\n}\n",
  "description": "The NYU-Depth V2 data set is comprised of video sequences from a variety of\nindoor scenes as recorded by both the RGB and Depth cameras from the\nMicrosoft Kinect.\n",
  "location": {
    "urls": [
      "https://cs.nyu.edu/~silberman/datasets/nyu_depth_v2.html"
    ]
  },
  "name": "nyu_depth_v2",
  "schema": {