Offered

Literature survey on word sense disambiguation (WSD) using WordNet will be done. In this project, Turkish WordNet will be used, thus the structure of Turkish WordNet will be studied in full. The existing WSD methods will be adapted to Turkish using Turkish WordNet and in addition a new method will be proposed.

Flakiness is an important problem in software testing. The goal is to develop a solution tor educe flakiness.

Example link:

https://engineering.fb.com/2020/12/10/developer-tools/probabilistic-flak...

In the last 3 years, we have collected a dataset of 25000 isolated sign videos belonging to ~650 different Turkish Sign Language gestures. These signs contain repetitions of the same sign from many different users. However, due to the scope of the data, it is hard to detect mistaken samples or classes from these videos one by one. Therefore in this project, we will implement an outlier detection application, that will analyze different characteristics of these videos to find mislabeled videos and correctly label them.

The project involves:

The project team will create a small database of famous paintings and their parodies. (Go to Google Images and query “arnolfini parody” to see what I mean).

Then, the team will develop image similarity measures to automatically detect parodies. Deep neural network and transfer learning approaches will be used. 

See the following paper for more information:

The project will involve writing AI bots for the slither.io game. 

Reinforcement learning and deep reinforcement learning approaches can be implemented eventually, but the first attempts will be based on heuristics. 

The bot will learn in two modes:

-Self play: many instances of the bot will play the game with different parameters, and the parameters will be adjusted based on the success of the bot.

-Learning by demonstration: a human will play the bot, and the game will be stored to learn from it.