You are on page 1of 2

lectures.quantecon.

org --> Phython language

openvim.com (Text Editor)

innovation@unibas.ch

Jupyter Notebooks

University of Basel Innovation Space


Gewerbestrasse 24, 3rd Floor
4123 Allschwil

Swiss Startups:
https://www.startup.ch/index.cfm?page=129571&branche_id=1&profilesEntry=1

https://www.fast.ai/

Tutorials on Kaggle

Innovation University of Basel/Startup Ideas


https://www.kaggle.com/learn/overview

https://www.unibas.ch/en/Innovation.html

U-Net is a standard for Data/Image Segmentation


Convolution Neural Networks

Pick up the competitions, and follow up the tutorials

Runtime is provided by Google


Kernels in Kaggel is the place where you will find solutions
Sort by Popularity
Baseline solutions and then improve the accuracy
We can create a different combination of different kernels
Enter the team during end of the competitions, so you can join the team

Competitions - House Pricing etc...

Data Augmentation solution


Notebook (Baseline Model) ipynb, like TGS_Salt.ipynb --> Save the file in google
drive --> Open using Colaboratory (Google environment, which provides GPU for free)
colab.research.google.com
Take a look at code
Runtime --> Change Runtime type --> Runtime type Python 3 and Hardware Accelerator
as GPU in the notebook settings

Kaggel usage? For initial validation of the code

Keras, pandas, numpy are Phython libraries, used for specific tasks

Insert > Code Cell

Kaggel API Key

Dataset from Kaggel

Training and Validation set of data

Spliting the data into training and validation sets


Train the model (.ipynb)

Visualize the training data

When you run, you will see RAM, DISK bars

Tensorflow v/s Keras???

Kaggel.JSON file use?

Kaggel -- Competition dataset?

Run --> CTRL + Enter

--------------------------
Import Libraries
Set parameters
Load the images
Train the model
Inference with the model

You might also like