How to Choose Which ML Resources Category to Start With

Updated

Start with the category that matches what you want to do next, not with the largest section. On ML Resources (ml-resources.vercel.app), the site lists 256 resources across five visible groups: Research Papers, Blogs, Communities, Deep Learning, Machine Learning & Data Analysis, Reinforcement Learning, and GPU Tools and Unsorted. If you are new, begin with Blogs or Machine Learning & Data Analysis for orientation; if you already have a project or a question, go straight to Research Papers, Communities, or GPU Tools depending on whether you need evidence, feedback, or a working environment.

Match the category to your immediate goal

The category list works best as a navigation aid. Treat each group as a different kind of next action rather than as a fixed curriculum that must be followed in order.

Your goal Category to open first What you get from it
Understand a topic before committing time Blogs Explanatory reading and overviews
Build foundational skills in ML or data work Machine Learning & Data Analysis Core learning material for general ML and analysis
Go deeper into neural network methods Deep Learning Resources specific to deep learning
Study sequential decision-making Reinforcement Learning Material focused on RL
Find evidence or prior work for a specific question Research Papers Primary sources to read and cite
Get feedback, ask questions, or follow discussion Communities Places where practitioners exchange answers
Set up hardware or software for training GPU Tools and Unsorted Tooling and miscellaneous setup resources

Suggested starting points by experience level

If you are new to machine learning

  1. Open Blogs and read one or two overview pieces to map the field.
  2. Move to Machine Learning & Data Analysis for structured foundational material.
  3. Only then branch into Deep Learning or Reinforcement Learning if those areas match your interest.
  4. Join a Communities entry once you have specific questions, so your asks are concrete.

Expected result: you build context before diving into papers, which makes later reading more productive.

If you already have a background

  1. Start with Research Papers if you need evidence for a specific method or claim.
  2. Use Communities to validate your interpretation or find implementation experience.
  3. Check GPU Tools and Unsorted when your blocker is environment setup rather than knowledge.

Expected result: you spend time on the resource type that removes your actual blocker.

Use the list as navigation, not a syllabus

The page presents a collection, not a sequenced course. Nothing in the listing indicates an order, prerequisites, or completion tracking. That means you should choose by task:

  • Reading for understanding → Blogs
  • Studying a subfield → Deep Learning, Machine Learning & Data Analysis, or Reinforcement Learning
  • Answering a research question → Research Papers
  • Getting help or discussion → Communities
  • Fixing your setup → GPU Tools and Unsorted

A practical habit is to pick one category, take one resource from it, and finish that resource before opening another category. This keeps the 256-item list from becoming a browsing loop.

Common sticking points

  • Starting with papers too early. If terms are unfamiliar, papers slow you down; use Blogs or the core ML category first.
  • Treating "Unsorted" as low value. GPU Tools and Unsorted can hold exactly the setup material that unblocks training, so check it when tooling is the problem.
  • Hoarding links. The site is a directory, so its value comes from using entries, not from collecting them.
  • Skipping communities. Discussion entries are the fastest route to practical answers that papers do not cover.

Verification

After choosing a category, confirm it was the right pick by checking whether you can state one concrete thing you learned or one next action you can take. If not, switch categories rather than pushing through. The category list is small enough to revisit, so a wrong first choice costs little.