2021 January

Documentation

Workflows

Tips

  • Click the beginning of the code block, then Shift+Click the end. That will select the entire block, which you can then copy normally.

Web Tools

Machine Learning

Machine Learning Course

Learning NN

Development

Hand Detection

Security

  • CRXcavator : Inspect Chrome Extension content for malicious code

Productivity

Machine Learning

Language

OCR

Language

Web Tools

Video

Audio Generator

Open Source Blog

AI Art

Traffic Simulator

CSS

Machine Learning

Decentralized

  • Skynet: The decentralized CDN and file sharing platform (like IPFS)

ebook

Neural Network

Paper

DNS

Course

Github Aps

Handshake DNS

Covid

IPv6 Check

Learning

Gmail

Javascript

Python

App List

Functional Web

Machine Learning

Dataset

  • EleutherAI EleutherAI is a grassroots AI research group aimed at democratizing and open sourcing AI research.
  • The Pile : The Pile is a 825 GiB diverse, open source language modelling data set that consists of 22 smaller, high-quality datasets combined together.

Machine Learning

Productivity

  1. At the beginning of the day, create one goal that I have to accomplish that day. The goal needs to be specific enough that I can definitively say when its done, and it needs to be completely realistic for me to finish it that day, so that I can say "Yes, I can definitely do this. No excuses". It also needs to be significant enough that at the end of the week, even if I only did those goals, I would still feel like Im making some consistent progress. Typically the goals I choose are things that might take one to two hours of focused effort.
  2. Only do one thing at a time, and log what I'm doing as I go. Before I start a task, I write down what Im about to do. When I finish the task, I check it off and write the next bullet point. If something comes up while executing the task, like realizing I have a question I need to ask someone, instead of asking them right away, I will write down a "TODO" in-line in my log and box it to come back to later. So if the task is "Check email", I write down "Check email", open my email, read it and execute only small tangential tasks (so, e.g., I dont see an email about a code review and then get distracted and go do the code review), and when Im done, I quit my email and check off my task. I do the same for Slack, which means I dont leave it open in the background. If I take a break, I write "Break" and check it off when I'm done and come back to my desk. I write my big goal at the top of the page and when I finish it, I box it and check it off for some extra satisfaction. At the end of the day I write DONE and box it, to get some closure. Note: I find if I write down 5 tasks and cross them off one by one, I will gravitate to the short easy ones first for instant gratification.
  3. Leave my phone in the other room while I'm working. It seems that I really do have to be physically separated from it by a significant distance to keep from being distracted by it.

Markup

Machine Learning Toolbox

Machine Learning Seminar

Machine Learning Notes

Fast.ai Online Course

Machine Learning

Pseudocode

Ensembling

  • Ensembling is the use of several independently trained models to form an overall prediction. The basic idea of ensembling is that individual models have weaknesses in different areas, which are compensated by the combination with predictions of other independently trained models. Possible ensembling strategies are e.g. majority voting, the use of a weighted average based on classifier confidences, or simply using the arithmetic mean of several predictions of different models and model architectures

Machine Learning Tools

Github Project Structure

Data Science Learning

Machine Learning Version Control

Generative Model Course

Paper Discussion

Course

Interactive Books

Internet Scanner

SVG

RSS to JSOn

Dynamic DNS

DNS

Generative Patterns

Hardware for Deep Learning

Podcast - Text to Speech

Learning

Programming Learning for Kids

Interactive Books

Style GAN

This Thing Does Not Exist

Style-GAN

Interesting Reading

NN Draw

Free Icons

Machine Learning Review

  • [Best deep CNN architectures and their principles: from AlexNet to EfficientNet AI Summer](https://theaisummer.com/cnn-architectures/)

Computer Science

Machine Learning

Ebook

Color

Desktop

  • Deskreen Deskreen is a free desktop app that makes any device with a web browser a second screen over WiFi. Deskreen can be used for mirror entire display to any device that has a web browser.

Python

Search Engine

  • Dash: search engine for gutenberg files

Image Editor

Food

About ML Books

If you're strong in software engineering, I recommend Machine Learning Mastery with Python by Jason Brownlee as it's very hands-on in Python and helps you run code to "see" how ML works.

If you're weak in software engineering and Python, I recommend A Whirlwind Tour Of Python by Jake VanderPlas, and its companion book Python Data Science Handbook.

If you're strong in architecting / product management, I recommend Building Machine Learning Powered Applications by Emmanuel Ameisen since it explains it more from an SDLC perspective, including things like scoping, design, development, testing, general software engineering best practices, collaboration, etc.

If you're weak in architecting / product management, I typically recommend User Story Mapping by Jeff Patton and Making Things Happen by Scott Berkun, which are both excellent how-tos with great examples to build on.

If you're strong in math, I recommend Understanding Machine Learning from Theory to Algorithm by Shalev-Shwartz and Ben-David, as it has all the mathematics for ML and actually has some pseudocode for implementation which helps bridge the gap into actual software development (the book's title is very accurate!)

For someone who is weak in the math of ML, I recommend Introduction to Statistical Learning by Hastie et al (along with the Python port of the code https://github.com/emredjan/ISL-python ) which I think does just enough hand holding to move someone from "did high school math 20 years ago" to "I understand what these hyperparameters are optimizing for."

Data Visualization

Microcontroller and Python

Math

E-learning Gamification

  1. Challenges
    These map to the learning objectives or learning goals.
  2. Levels
    These map to the learning journey and as the learner goes through each level, it signifies a step up in proficiency for them.
  3. Instant feedback
    This helps learners know how they are faring against their learning goals and based on this, they can adopt the necessary measures to step up their performance.
  4. Scores
    They are indicators of their performance and are closely aligned to offering gratification as well as a sense of accomplishment.
  5. Badges
    As the learners go through the learning path and clear certain levels, they are given badges. These reflect affirmations for their significant achievements.
  6. Leaderboards
    They are dashboards that are used to provide a pictorial view of the overall progress—including against others. The analytics keeps learners connected to the learning journey and aligned to meeting their terminal objectives.
  7. Competition
    This can be leveraged effectively as it helps learners assess where they stand against other peers or competing teams.
  8. Collaboration
    This feature not only facilitates team-building but also enables learners to leverage the support of peers or guidance from experts to meet their goals.
  9. Points and currency Points are a variation on badges, and can also be used as currency within a program. For instance, you can offer learners a point for every comment the leave in the discussion forum, and then let them redeem a certain number of points for rewards and discounts.
  10. Progress bars and level up Progress bars can help by visualizing their advancements. You can even add a bit of interest by creating "level ups" that unlock new course features.

Statistical Tools

  • Shiny Server Daniel Lakens: Recovering data from summary statistics, App to perform simulation based power analysis for ANOVA designs

Full Stack Deep Learning

Flight Dynamics

Ebook

NLP

Interactive Web

CNN Sliced Inference

obss/sahi: A vision library for performing sliced inference on large images/small objects

Deep Learning Noise Cancellation

Math Visualization

Windows App

LORA

Meshtastic An open source, off-grid, hiking, climbing, skiing, GPS mesh communicator. Similar to disaster.radio