Applied AI #6 Report

The sixth batch of our most coding packed program Applied AI Study Group was concluded, and it took place between January 15 and February 26. Our participants gained hands-on experience on popular computer science related topics such as natural language processing, computer vision, anomaly detection, predictive analytics for 6 weeks. 7 talented guides from inzva AI community made it possible for our participants to practice each week’s topic with dynamic real-life coding experience.

Our participants went through a strict elimination process that includes both theoretical questions and a coding question on HackerRank. 21 participants from 12 different universities graduated from this batch of the program after completing an immense 6-week period. They had a productive winter, meeting each other online every Saturday and doing 3 assignments throughout the program. For those who missed the application process, all the content and assignments are uploaded to our GitHub repository as open source to benefit all the AI community. 

Due to the increase in Covid-19 cases in January and February, we could not make our sleepovers in Kundura, but we were able to organize a face-to-face meetup and had a great time discussing the program and assignments. With the change of Covid regulations and the re-opening of schools in March, we invited our participants to the first AI sleepover of 2022! 

Throughout the 6 weeks, in addition to hands-on coding experience, our participants improved themselves by asking questions, getting feedback and taking advice from our both inspiring and encouraging guides Abdurrahman Dilmaç, Ahmet Melek, Duygu Ay, Enes Sadi Uysal, Gürkan Soykan, Halil Eralp Kocaş, and Şafak Bilici. 

Don't worry in case you missed Applied AI Study Group #6, you know inzva will not stop. Stay tuned for upcoming programs!

Our Guides:

Take a look at what we have learned during the program!

January 15: Computer Vision led by Halil Eralp Kocaş

We will work on image classification problems, object localization and object detection using original notebooks to gain hands-on experience related to computer vision. We will also go through some state-of-the-art architectures that are used to solve these problems.

January 22: Natural Language Processing Part 1 led by Ahmet Melek

We will make an introduction to the NLP field, covering tasks, problems and different ways of modelling. We will introduce tasks such as extractive QA, text classification, language generation, and entity recognition. We will implement and train language models, and work on text classification. Finally, we will be talking about various important concepts such as preprocessing and data formats, frameworks, and being up to date with SOTA research.

January 29: Natural Language Processing Part 2 led by Gürkan Soykan

We will continue to cover various problems such as neural machine translation, named entity recognition, and question answering. We will be implementing models from scratch, training them, utilizing popular frameworks to get predictions from trained (ready) models, and deploying ready-to-use models in a web interface.

February 5: Anomaly Detection, Hyperparameter Optimization, Recommender Systems & Time Series led by Enes Sadi Uysal and Şafak Bilici

We will focus on anomaly detection and imbalanced datasets. We will move on to various deep learning and machine learning models to perform anomaly detection, such as CatBoost, XGBoost, Isolation Forest, Autoencoders and TabNet. Then, we will study theoretical and practical aspects of Hyperparameter Optimization algorithms, such as Random Search, Grid Search and Bayesian Optimization. Lastly, we will study Recommender Systems using various algorithms and learn about time-series data and AI algorithms that process the data.

February 12: Predictive Analytics led by Abdurrahman Dilmaç

We will review the basics of machine learning by going over classical ML algorithms with scikit-learn. Then, we will learn how to determine a baseline when faced with a new problem and how to make sure your model is learning something. We will introduce MLFlow for experiment tracking and Ray Tune for distributed hyperparameter search for both classical and deep learning models. After talking about overcoming common data problems using NumPy, Pandas, and imblearn libraries, we will introduce ensemble models with Scikit-learn and LightGBM.

February 19: Deployment by Duygu Ay

In our final week, we aim to familiarize the participants with the ways to deploy machine learning and deep learning models to a production environment.  We will begin by learning how to set up a cloud server, then we will figure out the tricks to serve our model using Nginx, Uwsgi, and Flask.  We will talk about the capabilities and challenges of production systems. At the end of the week, participants will know all about sharing their work with the world.

HOW DID OUR PARTICIPANTS DO?

COMMUNITY FEEDBACK

“This was an incredible experience, especially for me, who had only worked on object detection and NLP before and whose practical knowledge was limited to projects he was working on. I knew there was much more to deep learning than I had discovered before this program, but Applied AI offered a completely different perspective. The sincerity of the community made each lesson much more enjoyable and at the same time, much more productive. In short, if your expectation is typical hands-on training, as I expected before, prepare yourself for much more. Assignments, notebooks, live lessons and homework; it was very unusual and enlightening compared to other hands-on programs. From the first minute of the first lesson, I realized that it would be much better than I expected... I look forward to working with this team again. Thanks inzva for the amazing experience."

“Although I had a background in the theoretical aspects of deep learning and AI from my academic studies, I have not previously been able to apply this knowledge to real-life applications. In Applied AI Study Group, inzva created the perfect environment for us to implement what was theoretical knowledge into practice. Throughout the program, I had the chance to work with different types of data such as language and vision. All the assignments given were well prepared for different levels of participants; they were highly educative, and they provided the right amount of challenge to keep me eager to learn and explore more. The lectures given by the guides provided me with the opportunity to fill in any gaps I might have between theory and application. The ongoing support of the guides throughout the program made the learning process much more enjoyable and smooth. After completing the program, I have much more confidence in my artificial intelligence skills. It was great to be part of such an amazing program. Thank you for everything!”

All participants have to abide by our CODE OF CONDUCT  and LETTER OF CONSENT. 

A  BEV Foundation project inzva is a non-profit hacker community organizing study and project groups as well as camps in the fields of AI and Algorithm; and gathering CS students, academics and professionals in Turkey.

Follow us on our social media accounts to have the most recent news about our upcoming events and programs!