I watch a lot of YouTube videos and have noticed that many content creators use transformed images to get around copyright laws. I personally only try to use copyright-free images in all of my posts, but some people prefer to use current, up-to-date photos. Creators such as Murky Meg, Celtic News, or others tend to use transformed images of Prince Harry and Meghan Markle to avoid litigation for using their images. In fact, the Christmas card that Prince Harry and Meghan Markle put out last Christmas in 2020, seen below, looked to me as if it had been transformed:-
I have been spe very steep learning curve, as compared to Python’s other machine learning library, sklearn. As datasets get larger and the data contained in them become more complex, it is imperative, therefore for any aspiring data scientist to acquire skills to be able to make predictions based on those larger and more complex datasets.
I have been studying convolutional neural networks, or CNN’s, and working on the CIFAR_10 dataset in particular. This dataset is composed 60,000 32 * 32 low resolution colour images of ten classes that allow data scientists to carry out machine learning research. …
In a recent post I wrote abIST dataset, as there are many ways that this can be done. I focused on the PyTorch and skorch libraries. PyTorch is a library that concentrates on machine learning and skorch is a wrapper for PyTorch that behaves like the sklearn library, which is a module I have more familiarity with. The link to this post can be found here:- Code Review: Google Colab MNIST prediction with sklearn and skorch | by Tracyrenee | CodeX | Apr, 2021 | Medium
In two earlier posts I discussed how PyTorch can be used to make predictions on the MNist datasets and was very pleased to have achieved a very high accuracy score using both the PyTorch and skorch libraries. The link for the most recent post on the subject can be found here:- How I increased accuracy 2% on Kaggle’s MNIST competition using PyTorch and skorch | by Tracyrenee | MLearning.ai | Apr, 2021 | Medium
In this post I intend to discuss how the PyTorch library can be used to make accurate predictions on the CIFAR_10 datasets.
The Canadian Institute for…
Just a few weeks ago the one year anniversary of the UK’s lockdown arrived. The lockdown was supposed to be for a few weeks to save the National Health Service (NHS), but a year later the UK has had to endure three lockdowns caused by surges in infections, as well as a more highly infectious strain arising out of London, England.
Just as the UK has hopes of getting to grips with the virus by enforcing lockdown measures and rolling out the vaccine across the nation, it appears that France is entering its third lockdown since the pandemic began. This…
In my previous post I carried out a code review of the solution to the openml’s MNIST dataset using PyTorch and a new library called skorch. The link for that post can be found here:- Code Review: Google Colab MNIST prediction with sklearn and skorch | by Tracyrenee | Apr, 2021 | Medium
For the past few weeks I have been endeavouring to improve my programming skills and have begun studying PyTorch, a library that spun off of the c++ library, Torch. PyTorch is Python friendly, so it can easily be integrated into a program written in Python.
One trying that I have been studying is convolutional neural networks, or CNN, and in particular I wanted to use a CNN in the MNIST dataset, which is a dataset that is composed of digital images of digits between 0 and 9. Because I am new, I researched the internet how CNN’s are used on…
PyTorch is a small part of a computer software that is based on the Torch library. It is a deep learning framework introduced in 2016, making it a new technology, comparatively speaking. The high level features provided by PyTorch are a Graphics Processing Unit, or GPU, and a deep neural network.
In my previous post I used PyTorch to make predictions on the complete Titanic dataset and found the accuracy achieved was no better than that achieved in any estimators of sklearn, statsmodel, or binary classifiers written from scratch. The link to my previous post on Titanic dataset can be…
I have been in the process of learning new programming skills and decided to take a free online course in PyTorch, which is an relatively new (circa 2016) open source library maintained by Facebook. I have tried Google’s TensorFlow in the past and I was never able to get the models to predict properly, so hoped PyTorch would give me better results. One thing I have learned from taking an online course in PyTorch is the fact that there are a lot of steps that must be completed before the estimator will give a prediction.
In order to improve my Python programming skills, I decided to take a free online Pytorch course. The course requires students to carry out lab exercises, and one such exercise was based on a small dataset concerning college admissions. The algorithm used in the course used a neural network written from scratch, which yielded an accuracy of 70%.
Since the student admission dataset is a small dataset, I decided to try it out on sklearn’s neural network and compare the two models, seeing which one afforded the highest accuracy.
The problem statement that accompanied this dataset reads as follows:-
I have over 46 years experience in the world of work, being in fast food, the military, business, non-profits, and the healthcare sector.