Intro to ML

What is in the Name?

AI, ML, DL....blah, blah


In [5]:
# Define them, with examples

Learning before ML Learning


In [7]:
# Maths, Stats, Programming

Application Domain


In [8]:
# NlP, CV, RecSys, 
# Finance, health, social network

Types of ML


In [3]:
# Discuss Supervised, unsupervised, labelled, unlabelled, reinforcement, indirect interpretation,
# Generalised, non-genralised, lazy/active, whitebox,blackbox

How do Machines Learn?


In [4]:
# Talk about error function, loss function
# ML =f(x), eg: series 2,4,6,8,?,?,  # finding the best fitting function
# Max IG in  Decision Tree, MSE in LR

ML Project Lifecycle


In [5]:
# Covered in previous session
# Explain the process of feature extraction, preprocessing, model selection, sklearn-workflow*(ext slide)

Sklearn/keras Workflow


In [15]:
"""
data.split()
model.fit()
model.evaluate()
model.predict()
model.save()
"""


Out[15]:
'\ndata.split()\nmodel.fit()\nmodel.evaluate()\nmodel.predict()\nmodel.save()\n'

ML in Production


In [6]:
# Talk about A/B testing, load testing, user feedback, offline training, online scoring

In [7]:
"""
model.load()
model_api.predict_real-time(user'input)
"""


Out[7]:
"\nmodel.load()\nmodel_api.predict_real-time(user'input)\n"

END