ryanholbrook

from joblib import delayed, Parallel
from pathlib import Path
from tqdm import tqdm

import hashlib
import imagehash
import pandas as pd
import PIL
import pybktree
from PIL import Image

ryanholbrook

import numpy as np

EPSILON = 1e-10


def _error(actual: np.ndarray, predicted: np.ndarray):
    """ Simple error """
    return actual - predicted

ryanholbrook

=
# +
import numpy as np
import matplotlib.pyplot as plt

plt.style.use("seaborn-whitegrid")
plt.rc('figure', autolayout=True)
plt.rc('axes',
       labelweight='bold',
       labelsize='large',

ryanholbrook

import tensorflow as tf
import tensorflow.experimental.numpy as tnp
from absl import logging
from typing import Callable, Dict, Union, Optional, Iterable, Sequence
from tensorflow import keras
from tensorflow.keras.optimizers.schedules import LearningRateSchedule


# Schedules ported from Optax
# https://github.com/deepmind/optax/blob/master/optax/_src/schedule.py

ryanholbrook

import math
from itertools import product

import matplotlib.pyplot as plt
import numpy as np
import tensorflow as tf
import tensorflow.keras as keras
from matplotlib import gridspec

# # Activation Model #

ryanholbrook

import numpy as np

import tensorflow as tf
from tensorflow import keras
from tensorflow.keras import layers

import matplotlib.pyplot as plt
from matplotlib import animation, rc
rc('animation', html='html5')
plt.style.use('seaborn-whitegrid')

ryanholbrook

Classifiers

Introduction

Looking at the decision boundary a classifier generates can give us some geometric intuition about the decision rule a classifier uses and how this decision rule changes as the classifier is trained on more data.

Plotting Functions

ryanholbrook

library(brms)
library(parsnip)

## Fits a brms model
convex_regression <- function(formula, data,
                              family = "gaussian",
                              alpha = 1, gamma = 2, # Yang (2014) recommends alpha=1, gamma=2
                              verbose = 0,
                              ...) {
  if (gamma <= 1) {