IoT Research Using Deep Learning

Tuesday, I read Amanda's 5G research paper on "A Review of Deep Learning in 5G Research: Channel Coding, Massive MIMO, Multiple Access, Resource Allocation, and Network Security". I also set up this website and started the tutorials on TensorFlow.

Wednesday, I completed a crash course video on YouTube for Deep Learning which lasted me a very long time. I also looked at another journal from IEEE Library titled "Intelligent IoT Connectivity: Deep Reinforcement Learning Approach".

Thursday I read another IEEE journal titled "A Deep Reinforcement Learning based Human behavior Prediction Approach in Smart Home Environments". I also looked at more Deep Learning information and concepts.

Friday I started taking a TensorFlow class on Udacity.

Monday and Tuesday I worked on my first neural network using the TensorFlow tutorial page. The program used the Fashion MNIST dataset to do basic image classification. I learned about overfitting data during preprocessing and read on ways to prevent overfitting.

Wednesday I looked into datasets, what makes up a dataset, how you can go about collecting your own dataset, and also looked into different websites that offer datasets

On Thursday and Friday, I did another neural network except from the TensorFlow tutorial page on text classification. I also read another journal titled "Deep Almond: A Deep Learning-based Virtual Assistant" from Stanford University that dove into how virtual assistants break down speech recognition.

On Monday and Tuesday, I spent most of my time creating my own dataset of pictures of clothing. I downloaded about 3000 pictures of different articles of clothing.

The rest of the week I spent trying to set up the dataset and replace it with the testing data that I used in the TensorFlow tutorial. I ran into many different problems however so my next goal was to try and use only one photo image and test the program.

I ran into trouble with this however and could not get the photo image to run with the program.

On Monday, I took a step back and worked on another tutorial from TensorFlow and crated a neural network to classify dogs, cats, automobiles, airplanes, etc..

Tuesday, I then tried to take three jpg images and use them as a subsitute for the testing dataset but again i ran into problems with the images. I believe the problem was with the format of the image but I was not too sure.

Wednesday and Thursday I spent working on a third tutorial. This tutorial was from the website datacamp.com and it did not use a keras dataset.

The program used many old functions and attributes that are no longer included in the tensorflow library so I have not been able to get the program to run currently.