Neural network from scratch: Part 1; gradient descent
Python project.
This article explains the principle of gradient descent and its implementation :
- Using the differential approach (2D example),
- Using the perturbation approach (3D example),
Python project.
This article explains the principle of gradient descent and its implementation :
Python project, Keras.
This article show how to use neural networks to detect the kindness of sentences:
Python project, Tensorflow.
This article shows how to get familiar with Tensorflow and how to use its great tool TensorBoard.
We will train a CNN on the MNIST dataset with few samples and show how to artificially increase our dataset (rotation, zoom, contrast, etc) to improve its accuracy. Then, we will explain how to implement a generative autoencoder to “dream up” new digits to improve our accuracy on real digits.
Continue reading “Data Aug. and Generative auto-encoder”Python and OpenCV project.
This article will show how to get the position of an object and its orientation thanks to an homemade Beacon.
Continue reading “Home-made Beacon”
Python and C++ projects.
To realize my bottle recycling robot project, I had to find a way to recognize bottles. This article will show how I installed a camera on the robot and the solutions I found to do visual recognition using: ImageAI lib, OpenCV and a Haar Cascade. Finally, how to convert the position of the bottles in pixels of the image into a real 3D position to guide the robot towards its goals.
Continue reading “Bottle recognition”
Python project, Keras.
This article will show how to get better results if we have few data:
1- Increasing the dataset artificially,
2- Transfer Learning: training a neural network which has been already trained for a similar task.
3- Unsupervised pre-training (if we have enough data but few have a label)
Continue reading “Transfer Learning & Unsupervised pre-training”
Python project.
Conway’s game of life, simple rules lead to incredible things.
Continue reading “Conway’s game of life”