NYP-AINYP AI Summer Camp 2020 is the first of its kind student-led bootcamp for Artificial Intelligence.
Through 4 days of accelerated learning of AI, students learnt an immense amount of AI theory and implementation. Students went from being complete beginners to being able to Predict House Prices, Preprocess Unclean datasets and even classify Spam messages. Delving into the fields of prediction, classification and Natural Language Processing (NLP), it proved to be a tough yet fulfilling journey for many students.
Day 1: AI Essentials + Regression
Being the first day of our Summer Camp, we had to ensure students had a solid understanding of Machine Learning, which would aid in their learning for the next few days.
Students were introduced to popular Machine Libraries in Python, namely: Scikit-learn, numpy and pandas. They were taught how to implement these libraries using python code.
As students were being introduced to predictive algorithms using Machine Learning, they were taught about the inner workings of Linear Regression, a popular Regressional Algorithm....
...And went on to predict housing prices using their newly acquired skills.
Day 2: Image Classification
Students were taught about one of the most popular classification algorithms: Logsitic Regression.
They went on to learn how to preprocess images, converting images into an array of numbers.
Having learnt the basics of Image Preprocessing, we went on to preprocess our own weather dataset, which consisted of 4 different weather types. Using this preprocessed dataset, we went to train our own weather classifier using Logsitic Regression.
Day 3: Data Pre-Processing
Unfortunately, Real-World Data isn't as clean as we wish them to be. Students learnt how to deal with Dataset Problems, namely: Missing Values and Categorical Values.
Students then went on to preprocess the Titanic dataset, a popular competition hosted on Kaggle.
Want to use a different algorithm? No problem!
Day 4: NLP & Fine Tuning
How would be let our Machine Learning Algorithm learn for text? Tokenization and Count Vectorizer, here we come...
Putting all these pre-processing to use, we created a text classifer using Logistic Regression, which would classify tweets as Disaster or Not Disaster.
Let's talk about hyperparameters, parameters which affect our model's performance...
With so many parameters to tune, how will we find the optimal combinations?
Day 5: Spam Classification Competition
What's a bootcamp without a competition to end it off?
Students went to form teams of up to 4. This 2 day competition saw participants using the skills acquired from the past 4 days: Pre-processing, NLP and classification.
NYP AI Summer Camp 2020 was indeed a new experience for all of us. We're extremely glad to have this opportunity to organize such an event, to have the privilege to teach over 70 students about AI.
We're always exploring new aspects of AI to teach, always looking for opportunities to push for the widespread education of AI. Do keep a lookout for our future events!
If you are interested in potential collaborations with NYP AI, we'd like to hear from you: https://about.nyp.ai/collaborate :)
Till then~
