- Today
- Total
Running Deeper C/C++/Java
Week 29 Report Main Content: 2 Rough Image & 1 3D Modeling ▶ Overall :: Enchanter :: Gryphon's Nest :: Enchanter 3D Modeling ▶ Development Status :: Map design and production finished :: Buildings 90% finished :: Units 40% finished :: Unit animations 10% finished :: UI Design started :: System 35% finished :: Server preparations :: Deep Learning preparations
DIY Electric Scooter Project: Reasoning: 1. Dorm to Soccer Field: 20 minutes walking, 10 minutes running 2. Dorm to Soccer Field: 10 minutes Kick Scooter, 5 minutes Biking 3. Time in between classes: 5 Minutes Thus, I needed something that would help me move faster. I owned a big wheeled Razor A5 Lux Red Scooter, and spare handlebars from the Razor A5 Lux Blue Scooter. Motor: 32V 20 A Controller..
Today, I tried writing a Generative Adversarial Network. Using the MNIST Data set, I will be writing a GAN that will create images of numbers as close to the dataset as possible. Currently, I have been getting stuck, the generator doesn't generate proper images. GAN works with two neural nets, one was a generator, the other being a discriminant. The generator creates images from random noise, th..
A convolution is a k*k grid of an image, that the CNN can read and analyze. Unlike a normal neural network, a convolutional neural network can incorporate the 2D style of the original image, resulting in a much higher accuracy than a normal neural network. Below is a nice explanation from Medium.org. Convolution 2. Convolving an image with a filter We take the 5*5*3 filter and slide it over the ..
Using the MNIST data set, I will be trying to detect numbers based on images, and get the accuracy as high as possible. The first step: the normal Flattened Neural Network. This method is much faster, but also less accurate, since the image's 2D properties turn into 1D, and lose the 2D shape. This causes a drastic loss in accuracy. The neural net I will be using is a Multi Layer Perceptron neura..
Above is another team from Phillips Exeter Academy that made it to Global Demo Day, an app that partners with shopping markets, that targets expiration dates for foods and recipes for better cooking, reducing food waste and creating a healthier/tastier eating experience. Global Demo Day, First Day: Convention, meeting the other teams The Recyclotron app, fully finished and on the Play Store Our ..