MITFutureMakers

Day 1 7.6.21

Prompt: What are you excited to learn about in the program?

I am excited to get more familiar with the different machine learning models as well as learning about natural language processing or image processing.

Day 2 7.7.21

Prompt: What did you learn today?

Today, with the help of Dr. Kong, I got to learn about everyone's inspring origin stories. I also learned how to make my personal narrative be told in a meaningful way. The structure of Challenge, Choice, and Outcome helped me reflect on different periods of my life that I did not think had a lot of influence to the person I am today.

Day 3 7.8.21

Promots: 1) What is the difference between supervised and unsupervised learning?

2) Describe why the following statement is FALSE: Scikit-Learn has the power to visualize data without a Graphviz, Pandas, or other data analysis libraries.

1) Supervised learning requires that the input data be organized already while unsupervised learning will learn which are the input data points and thwir coorelation. Supervised learning gets trained by you while unsupervised learing it gets trained by past experiences.

2) Scikit-learn is created on top of all the other analysis libraries. It is false because scikit-learn requires graphviz to visualize the ML models in the scikit-learn library.

Day 4 7.9.21

Promots: Find a dataset

https://www.kaggle.com/ajaypalsinghlo/world-happiness-report-2021 //How can we make sure the world is happy using policies?

https://www.kaggle.com/datasciencetool/covid19-vaccine-tweets-with-sentiment-annotation //Use this data to prevent covid-19 vaccine misinformation?

https://www.kaggle.com/miroslavsabo/young-people-survey //Identifying different phobias based on gender and interests?

https://www.kaggle.com/geomack/spotifyclassification //improve spotify reccomendation system?

Day 7 7.12.21

Prompts: 1. What are “Tensors” and what are they used for in Machine Learning? 2. What did you notice about the computations that you ran in the TensorFlow programs (i.e. interactive models) in the tutorial?

1. Tensors are multidimensional arrays that can hold data such as integers and floating point numbers, and strings. You can think of this as matrix of numbers in your math class! You can use this to store images as numerical values for ANN.

2. I noticed that images with the color red and had simple shapes were ranked higher in the list of images.

Day 10 7.15.21

Prompts: 1. How do you think Machine Learning and AI were used in the design of the game? 2. Can you give a real-world example of a biased machine learning model, and share your ideas on how you make this model more fair, inclusive, and equitable? Please reflect on why you selected this specific biased model.

1. At the beginning of the game, ML and AI were implemented by collecting our inputs as sample data for the ML model to learn from.

2. In 2015, Amazon's hiring process used AI. It was later discovered that the algorithm favored men over women applicants in the process. This was due to the sample data used to train the model where the majority were men applying to work at Amazon. A fix for this is to use a balances dataset with equal amounts of gender. This type of model is dangerous and can leave perfectly capable applicants with zero chance of getting the job simply because of their gender instead of being valued for their skills. This also makes the gender gap in the CS/Tech industry bigger.

Day 11 7.16.21

Prompts: Succinctly list the differences between a Convolutional Neural Network and a Fully Connected Neural Network. Discuss layers and their role, and applications of each of the two types of architectures

In fully connected Neural Network, all neurons are connected to each other in each layer and each layer is connected to each other. Which means the input type can be anything. In CNN, because we know the inputs will be images.

Day 15 7.20.21

Prompts: Write a reflection piece on the advantages of the Rectified Linear activation function, along with one use case.

The ReLU activitation function only requires to use a max() function as opposed to using an exponential function, making it easy to implement in your notebook. ReLU also takes negative inputs as true zero values which makes classification problems easy to process while sigmoid and tanh activation functions have to learn how to approximate to true zero values. It is reccomended that ReLU be the default activation function for CNN models and not RNN models.