.md .pdf

repository

Contents

Introduction

So how should we start?

I have already mentioned I want to get a lay of land to motivate myself to begin my journey into labriynth of algorithms, ideas and applications. I am infact not a novice in machine learning. I have executed big projects, delivered end to end solutions; I am by all account an expert coder. What is the most efficient path that I should take? Well I need to capitalize on the knowledge and skills I already have acquired from years of experience; leverage them to best of my abilities and make my own way forward.

Well this is all fine but what is my end goal? I can imagine a few areas in my personal life and my work which can greatly benefit from application of NLP.

• Data mining knowledge base and yammer.

• Using NLP to review, summarize and generate insights on DDRs; Safety, event or incidence histories.

• Text mining and web scraping Oil & Gas knowledge bases, publications (papers)

• Social network analysis like: Facebook, Google, Whatsapp

With advent of GPT-3, there are some interesting things happening in the world which I plan to learn and attack eventually. For now they serve as a source of inspiration of what is actually possible.

• Top 10 experiments in GPT3

• Seven online funtools created with GPT3

• Open AI api

• Twitter

• AI Dragon Model

• GPT3 Code Generator

What I am going to do?

This is a very high level game plan

Read News and Articles

I will continue to read on inspiring examples like above and keep building on a repository of intuition and ideas

Collate Advice on topics to cover

https://www.kaggle.com/c/jigsaw-unintended-bias-in-toxicity-classification/discussion/90291#latest-523764

Blogs, Courses and Learning Resources

• Blogs

  • natural-language-processing-is-fun

  • Sebastian Ruder Blog

  • Recommended blog list

  • Unreasonable progress of DNN in NLP

  • Big Bad NLP Datasets

  • Super Duper NLP Repo

• Kaggle Kernels

  • Approaching almost any nlp problem

  • Spooky Author Classification

  • Deep learning for NLP:Zero to transformer and bert

• Courses

  • Fastai:Course2020-Practical Deep Learning for Coders

  • Fastai:Code First NLP

  • Kaggle:Introduction to NLP

• Books

  • Approaching (Almost) Any Machine Learning

  • Deep Learning for Coders with fastai and PyTorch: AI Applications Without a PhD

Tools

• NLP Libraries

  • nltk, spacy, fastext

  • scikit_learn

  • fastai-v2

  • huggingface - nlp, transformer

  • huggingface+fastai-v2 - Hugdatafest, fasthugs

  • keras+tensorflow

• NLP Specific

  • Google LIT

• Visualization

  • Spotfire

  • PowerBI

  • Dash

  • Viola

  • D3

  • streamlit

• MLOps

  • mlflow

  • kubeflow

  • dvc

Learning Goals

• 5th September, 2020 (Beginning)

  • Learn basics of NLP and apply knowledge into couple of kaggle datasets

  • Learn enough about BERT to write a couple of patents by next 2 months on topics like Drilling & Oil and Gas Knowledge Management etc…

Welcome to Code First NLP Lay of Land - Spook Author Identification