Deep learning has taken the world by a storm in recent years. From self-driving cars to predictive advertising, it has inevitably become a major part of our day-to-day lives.
Geoffrey Hinton, the person credited to bringing deep learning back into the mainstream and largely responsible for what it is today, released an entirely new type of neural network known as the Capsule Network (or CapsNet). As the name suggests, it uses capsules rather than artificial neurons for its activities.
In this piece, we’ll build a deep learning model to classify objects in an image. To build the convolutional neural network, we’ll use this dataset available at Kaggle. A CNN is a type of neural network primarily used in visual tasks. The network will detect the features of an animal, and then use those features to classify a given input image as either a cat or a dog.
Continue reading Build a deep learning model to classify images using Keras and TensorFlow 2.0
In this post, we’re going to unravel the mathematics behind a very famous, robust, and versatile machine learning algorithm: support vector machines. We’ll also gain insight on relevant terms like kernel tricks, support vectors, cost functions for SVM, etc.
A support vector machine (SVM) is a supervised machine learning algorithm that can be used for both classification and regression tasks. In SVM, we plot data points as points in an n-dimensional space (n being the number of features you have) with the value of each feature being the value of a particular coordinate.
Continue reading Understanding the Mathematics behind Support Vector Machines
Tyleeseeuh seems to have endless lenses. And endless creativity. I went from a sunflower field to a virtual beach vacation before I’d finished my coffee (while wearing a crown of dinosaurs, of course). The creator said she uses lenses as her art form — from makeup looks not possible in reality to neon filters that mimic a sunset. She’s spent the past two years creating all sorts of lenses and filters and I managed to get a firsthand account of her journey.
Continue reading Snapchat Lens Creator Spotlight: Tyleeseeuh
Piotar Boa is probably the man behind your favorite lenses. Known for his vast clientele list and innovative AR lenses, he’s a Lens Creator and Spark AR Partner and has worked with numerous brands to bring new AR experiences to millions of viewers. We caught up with the busy Italian to learn more about how he views the industry and what he hopes to see in the future.
Continue reading Snapchat Lens Creator Spotlight: Piotar Boa
When training an artificial neural network (ANN), there are a number of hyperparameters to select, including the number of hidden layers, the number of hidden neurons per each hidden layer, the learning rate, and a regularization parameter. Creating the optimal mix from such hyperparameters is a challenging task.
Continue reading Selecting the Best Architecture for Artificial Neural Networks
Running deep learning models is computationally expensive. And when it comes to image processing with computer vision, the first thing that comes to mind is high-end GPUs—think the 1080ti and now the 2080ti.
But it’s hard to run computer vision models on edge devices like Raspberry Pi, and making a portable solution is difficult with deep learning libraries like TensorFlow or PyTorch.
Continue reading Real-Time Object Detection on Raspberry Pi Using OpenCV DNN
In this article, we’ll introduce the reader to Generative Adversarial Networks (GANs). We assume the reader has some prior experience with neural networks, such as artificial neural networks.
Here’s the plan of attack:
Below is a conversation about python class basics. To help follow along, you may need to understand Python functions.
Anita: Hi DarkAnita. I’ve been trying to improve my Python skills by using Python classes, but it just looks so much like gibberish to me. Python functions, I totally understand those…but classes? Do you understand them?
A good data scientist is not one who knows all the fancy algorithms but one who knows that he/she is overfitting. We all have been through that time when our super awesome, fully tuned model has failed to live up to the expectations on Kaggle private LB or after deployment. Knowing how to get an unbiased estimate of the predictive power of our model is important. There are different validation strategies like holdout and cross validation which are commonly used in practice for this. But which strategy is appropriate in which scenario is something that needs more discussion and thought.
Continue reading Machine Learning Model Evaluation & Selection
