Hole Enter.jpg “Where shall I begin?” The White Rabbit asked. “Begin at the beginning” the King said, “and go on till you come to the end: then stop.” — Alice’s Adventures in Wonderland

This page is for the curious (and curiouser). It is designed to be irreverent and unofficial, with lots of idioms and cultural references. The 'Rabbit Hole' concept comes from Alice in Wonderland, where Alice falls down a rabbit hole encountering many wondrous things. Thus, this is a collection of interesting and wondrous links, which are somehow related to Machine Learning tools.

Warning content does not have to be on the curriculum of COMP309, e.g. content more relevant to COMP307 could be signposted if it's interesting.

It will contain links and ideas useful for solving assignment tasks. At 300 level, explicit recipes on how to solve assignments are inappropriate as students need to cultivate their own problem-solving skills.

Tools

Basic tools

WYSIWYG editors suitable for assignment 1:

Weka Easiest place to start for the majority of tribes

KEEL Useful for the EC Tribe

Intermediate tools

KNIME Highly rated platform with clear pipelines

Scikit-learn Scikit-learn is an open source machine learning library for the Python programming language

Advanced tools

Keras for TensorFlow Keras is an API designed for human beings, not machines. Create your own pipelines using Python for TensorFlow

Blogs and Lists

Kdnuggets 25 years of topical knowledge discovery in databases
, e.g. Intro to ML {Lect1},
Visualisation {lect11},
Begin NNs here,
Feature engineering,
Maths for ML {Assignment 2}
Continuing beyond this course
Job roles,
Mastering ML in Python

How to use feature selection in Weka with walkthrough here {Lect8}

Data-mining blog An interesting collection of links and resources for R and Python.

TPOT Blog by Randy Olsen. Automated pipelines.

Regression

Reinforcement-learning plus game ai

Difference Between Deep Learning And Reinforcement Learning?

Data visualisation zoo covering a variety of interesting methods.

Meaning in ML key article by a key researcher.

Google AI, applications for good

MOOCs and Courses tools

Official Weka MOOC with excellent videos, such as creating pipelines with KnowledgeFlow. Long if you wish to explore all parts, so best used when needed.

KnowledgeFlow

Wrapper Feature Selection

Append and Merge

Scikit video Basic, but might help.

Pattern recognition slides Not basic, but interesting. Very good Computer Science.

Receiver Operating Characteristic curves: excellent video on ROCs and AUCs {Lect10}.

Deep Learning introduction plus links,
intuitive explanation of CNNs and overview of GANs.

Deep learning history

Repositories

Data, methodologies, algorithms and competitions:

Data

UCI ML repository

NZ Data respository

Google's new Dataset finding search engine{Assignment 2}

Medical datasets

Kaggle: fruits360

benchmarks ai

Methodologies

Crisp-dm.pdf: Crisp-dm comprehensive overview, also available from here

CRISP-DM Slideshare

Algorithms

Clustering

Competitions

Kaggle Competitions and interesting datasets

5 Tribes overview and other non-AI articles

The data science Venn Diagram where other variations are plausible.

The Essential Guide to Training Data: company spin, but interesting.

Coding and GPU Help

R:

R Data Exploring

Python: 10 useful Python resources

Python set-up

How to set-up coding

python_data_wrangling

Basic Stats

pandas-cheat-sheet-python

10-coding-mistakes-data-scientists

GPUs:

Google Colab {Ass4}
Video tutorial

Deep learning 101

Data Visualisation:
Luna Near idea for a visual data processing language, although have not personally tried it.

Research papers and books

Comparison of Decision Trees in Weka. Not the most advanced/best paper, but could provide an easy introduction to the topic.

An Excellent machine learning introduction for computer scientists

The book on Deep learning.

The link to the book on RL: www.cs.ualberta.ca/~sutton/book/the-book.html Note, this link is now obsolete. Here is an alternative link

Symbolic vs Connectionist AI.

Counterfactual reasoning the computer science behind ad placement in the Bing search engine.

Machine Learning: The High-Interest Credit Card of Technical Debt. An agile take on ML

machine_learning.pdf: Introductory machine learning book, also available from here

Murphy book on Machine Learning. A hard resource suitable for those who wish to go on to COMP421. Matlab code in links. A pdf of an earlier edition was available, so worth hunting down.

UCI Algorithms A comparison of classification algorithms on UCI datasets

Supervised and Unsupervised Discretization of Continuous Features

Feature selection and variables{lect9}
---++ Fun stuff

Short videos on AI

Electric Sheep

Heat mapping with Deep NNs

Comic Big Data discussions

Fake people

AI blog

And yet, most machine-learning systems don’t uncover causal mechanisms. They are statistical-correlation engines.


DSDN 487 Special Topic: Creative Artificial Intelligence.