How to use deep learning to predict the future
Is it possible to predict tomorrow’s Bitcoin price? Or if that’s too far a leap, what about the price 20 minutes from now?
I’ll be using the Long Short-Term Memory (LSTM) RNN machine learning model to predict the Bitcoin price 20 minutes from now, relying solely on simple historical financial data.
I’ve written this article partly as a guide, and partly as an exercise exploring the potential use of the LSTM model for the purpose of Bitcoin price prediction. Hence I may skip over some of the fundamentals, as these are easily found elsewhere.
A disclaimer that my experience comes primarily from curiosity, practical applications at a personal and professional capacity, and a Fama’esque intrigue into efficient markets. This may mean some of the terminology and methodology used could differ from others. …
For anyone not interested in learning how to build an application that watermarks your videos and just wants a simple way to add a watermark to a video; this utility will do just that!
Watermarking first appeared in Italy during the 13th century, where paper manufacturers changed the thickness the paper whilst it was still wet. The watermarked part of the paper let through more light leading to a mark that allowed readers to identify where the paper was produced.
Today we use watermarks for a wide variety of applications, with the majority of watermarks now being off the digital variety. It provides for a clear, yet relatively unobtrusive way, to show original authorship. This is particularly important in the age of the internet where it is easy to copy and appropriate media without permission. …
Using JSON to create beautiful videos
Video editing has never been something I truly enjoyed. I often found video editors cumbersome and wasn’t willing to spend the time learning to use professional video editing applications, let alone pay hundreds of dollars per year on an Adobe subscription. I did always enjoy programming, primarily because of the flexibility it afforded me to create whatever I wanted; bound only by my imagination and my skills as a developer.
The video shown above has been created with nothing other than code. No editing software has been used to design or render this video, with only a 40 line piece of JSON being needed to put together all individual elements to create a fully composited video using the Shotstack API. This API allows you to describe your video edit in JSON, and then use your favourite programming language to render hundreds to thousands of data-driven videos concurrently in the cloud. …
This is a continuation of our initial simple OTA solution for the ESP8266 or ESP32. In this next part we’ll look into how we can build a more comprehensive OTA solution that does the following:
- Build the ESP8266 into something that is actually half useful; a humidity, temperature and moisture plant sensor.
- ‘Dial home’ to check whether there are new firmware updates available.
- Update the module depending to the latest version number as defined by our database.
I have found this a very useful thing to learn while I’ve been trying to reduce the size of my projects through the use of surface mounted ICs. While they can be incredibly small, it’s pretty much impossible to reprogram them after soldering them to a PCB! …
This guide explains how to build a simple ESP8266/32 firmware updater in NodeJS.
This is the first part of a two-part guide. If you already know how the OTA update works and want to build an automated firmware version controller you can skip ahead to Part 2 here.
The ESP8266 will send a GET request to my application with some key information stored in its header. This will then be used to serve up the appropriate binary.
For this example I’m using the ESP8266 ESP12E.
