- languageSummary file created with a summary of Python syntax.
- fixed markdown lint warnings.
- I found Neetcode 150, a list of algorithmic coding problems based on the famous Blind 75 list. I will try to use as programming language Python to have some practice and go over as many of the 150 problems from the last as possible.
- First chapter I started is Arrays & Hashing
- Learned about web scrapping with Selenium and a webdriver for chrome. This can be done in a headless environment, like the Ubuntu server where I do most of my development but best is to run it in a desktop environment.
- First concept in chapter 4 is about mean squared error. It's a good thing to square an error, because it makes small errors smaller and big errors bigger.
- page 54
- The book makes reference to NumPy. I started to read the User Guide available on the website to try and get a quick overview of what the library offers.
- The chapter starts to become difficult to follow. I decided to go trough it at a faster rate without focusing so much on taking detailed notes. Hopefully I will be able to begin building a mental image of neural networks then come back to certain chapters for details.
- Chapter 3 which goes into detail about forward propagation is done.
- I need to go over a few examples of NumPy dot function. I found an easy to understand explanation of dot function on matrices here which is worth to bookmark for the future.
- page 48(started chapter 4)
- Learned about weighted sum network
- The interface for a neural network is simple. It accepts as inputs a list of information and a weights as knowledge. It outputs a prediction based on the inpus.
- Page 34
- The neural network outputs a prediction based on the input and the weight.
- Th network can give false predictions then it should try to adjust the weights in order to predict more accurately the next time it sees the same input
- Implementing a simple neural network with a single node
- Network sensitivity is related to the weight. If the weight is very high then even the tiniest input can create a large variation in prediction.
- page 27
- Created support for solving challenged from Hackarrank.com in python. No solutions yet added to the repo
- I found an interesting series on bdtechtalks called AI education. First article in the series contains a promising approach in self studying Deep Learning using Andrew Trusk Grokking Deep Learning book. The article mentions that this book doesn't need a strong mathematical background in linear algebra and calculus to start learning this field. I decided to give the book a try and see how long before I loose interest in it.
- So far the book promises to provide an easy approach in teaching what deep learning is which will prepare me for learning one of the existing frameworks(Torch, TensorFlow, Keras and others)
- 90% of the projects I start, something doesn't work from the first step. In this case, I get a "The folder you are executing pip from can no longer be found." when trying to install numpy with pip. I guess I'm going to start debugging this and by the time I'm done with it I will need a break from the mental exhaustion :(
- That was easy :) Apparently I run the commands from a directory that I deleted and that's why I was getting the error above. Next step is to figure out if I need Jupyter Notebook or I can code on my code-server instance in plain Python without additional support.
- It takes too much time to understand the benefits of Jupyter Notebook so I decided to not use it for now. Hopefully I won't regret it later on. Finally I can get back to the book.
- Deep learning is a subset of methods in machine learning toolbox, primarily using artificial neural networks inspired by the human brain
- Supervised vs unsupervised learning -> this is about the type of pattern being learned
- Supervised learning is the direct imitation of a pattern between two datasets. What you know -> supervised learning -> what you want to know
- Unsupervised learning transforms one data set into another, which is not previously known or understood
- Parametric vs nonparametric -> this is about the method for learning
- Parametric learning is characterized by having a fixed number of parameters
- Nonparametric model is characterized by having an infinite number of parameters, determined by data
- Page 44
- Reworked the existing python files because they were not compiling with with python 3
- Practiced a bit Markdown syntax by reworking the README.md file in the repository in order to keep a status of activities done. This way it's easier to remember where I left in case I need to take a break and resume this activity later on. Found a great reference online for quickly checking different syntax elements
- Started going trough Automate the Boring Stuff with Python chapter 8
- Continue going trough Automate the Boring Stuff with Python chapter 8
- Found a new repo on GitHub name Practical Python which seems to be worth going trough since it offers a more compact organization of information. Decided to set on hold the current course and continue with this one.
- Organized the files in the repo based on the different course or library that I study. So far all the scripts are split into 3 categories:
- Automate the Boring Stuff with Python - contains scripts created while going trough the Automate the Boring Stuff with Python course
- Others - contains misc scripts from unknown courses
- PyGame - contains scripts used for learning the PyGame library
- Practical Python Programming - contains scripts created while going trough the Practical Python Programming course from GitHub