=======================

30 Days CODING challenge

For the past few months I have been learning Deep learning and Artificial intelligence. Most of my work till now has been in Python. And as much as I like to stay focused on a few libraries related to AI and ML only, I found that not a sustainable way to go.

There are few key things like Data handling, Data cleaning , web apps to deploy the model;  that need to be mastered for a good overall impact.

So my Aim would be to have an overall holistic skill for getting things done in Python.

Main idea – Learning basic skills to supplement the AI/Ml skill, and at the same time, not making it boring.

The reason why I am sharing it with all is to keep track of my progress, and meanwhile it might motivate some of you to work on something similar.

Below is my list of 30 days projects, with defined goals to achieve. I have intentionally kept a break for 2 days, to cover up for any un-eventup delays.

CODING 30 days challenge(Starts 7 june,2020)

1. 1. Live Stock monitoring Web app
      • Goal – access to stock values of 10 desired companies in python and web app
   2. Stock market prediction web app
      • Goal – ML approach to predict rise or fall of prices- in next 1 day,1 hour?
   3. Enhancing DICOM images
      • Goal – access DICOM file from python, Visualizing in python without down scaling
   4. Importing NIFTI dataset
      • Goal – access NIFTI file from python, Visualizing in python without down scaling
   5. Training a 3D CNN
      • Goal – Train 3d-CNN on 3D MRI NIFTI dataset
   6. Create a “editable google forms” like – web app
      • Goal – understand data distribution, well versed with data handling, coding without copy paste
   7. Importing 2+6 diff types(from folder, csv, 1-data frame, 2-  from_name_re 3-from_name_func    4-from_lists  5-create_from_ll    6-single_from_classes)
      • Goal – understand data distribution
   8. Importing 5 datasets(different from number 7) in less than 10 minutes/dataset
      • Goal – reduce time from data import to training model
   9. “Exploring dataset” notebook for 5 datasets Kaggle
      • Goal – understand data distribution, well versed with data handling, coding without copy paste
   10. Train Cartpole – RL openAI in google colab * Goal – understand pipeline of OpenAi environment, visuals in Colab notebooks, Agent that can balance beam
   11. ——————————————Break————————————————-
   12. Display heatmaps in Fastai CNN web app * Goal – Display input image, heatmap, prediction class and probability in web app
   13. Train a GAN autoencoder * Goal – Theory part of autoencoder, generate fake Xray dataset?
   14. Generating Bounding Box(rectangle) in CNN results * Goal – training on UNet? Draw bounding boxes for segmentation results
   15. Generating lung border in X Ray * Goal – training on UNet? Draw outline for segmentation results
   16. Generating my own segmentation(box) dataset * Goal – Using Online tools, Bounding boxes for ventricles in MRI scans
   17. Generating complex shaped segmentation dataset * Goal – Online tools for segmentations? On Ipad? Detecting ventricle shapes on MRI
   18. Drawing ECG image from signal * Goal – understanding ways ECG data is stores, importing ECG signal, draw ECG
   19. Draw and detect characters – web app * Goal – Classifier on MNIST dataset, Webapp that support drawing and then gives predictions
   20. Image recognition from live Video * Goal – understanding CNN for Video, simple face detection from web cam
   21. ——————————————Break————————————————-
   22. OpenAi Algorithm environments – Reinforcement learning(RL) * Goal – understand data distribution, well versed with data handling, coding without copy paste
   23. Training RL Pac-man * Goal – understand Pac-man environment, Agent that can play game decently
   24. Training RL Atari game * Goal – understand Atari environment, Agent that can play game decently
   25. Training Robotic arm – RL openAI * Goal – understand Robotic arm environment, Agent that can pick up objects
   26. Training Mujuco – walking man OpenAI * Goal -understand Mujuco arm environment, Agent that can walk properly
   27. Training GymGo – RL OpenAi * Goal – rules of GO, select 3 most basic moves, train an agent that can do the most basic moves.
   28. Virtual “Old-data based Stock make Data generator * Goal – different ways to generate fake data that makes sense, generate data for a assumed xyz company
   29. Virtual AI investor * Goal -generating fake currency, make predictions, ask it to invest, test how it performs
   30. Creating your own RL environment * Goal – plan simple rules and rewards for the environment, Train a Agent that can do the thing

I have decided primary and secondary goals for all of them. Aiming to achieve as much as possible.

Finally, i love talking to smart and interesting people. So, if you wish to talk or collaborate, please feel free to ping me. Whatever way you prefer.