DBIS

This thesis aims to explore Deep Reinforcement Learning based methods for strategy optimization in the Contract Bridge game, such as Multi-agent Reinforcement Learning, Value-based Learning, and Policy-based Learning, as well as Knowledge Representation and Reasoning based methods such as Multi-agent Epistemic Situation Calculus

Edge computing offers a framework for offloading the computational effort required by IoT applications.
However, privacy issues and cost constraints for cloud infrastructures pose substantial difficulties to IoT service deployment. We wish to look at the possibilities of Peer-to-Peer networks and WebAssembly for IoT task offloading. As a result, a comparison of existing offloading tactics and information structures based on peer-to-peer and client-side solutions employing modern Web protocols such as WebRTC and WebAssembly is made.

The aim of this thesis is to build a MLOps-based system to compare the variations in the motions executed while performing various sports-related movements. In essence, we are interested in comparing movements across multiple sensor-suit based recording sessions.

Politicians are highly public individuals. A lot of data on them is available online. Many try to use social media such as Twitter to express their views, support other members of their party or interest groups, and interact with the public. Some also use it to branch out and connect to other countries’ politicians or respond to international events. With all of this, they form social graphs with them as nodes and their interactions as edges. A politician’s job, however, also often comes with balancing different interests, especially in times of uncertainty; this may lead to changing opinions based on the current state of events. However, as they try to influence the public on social media, other actors may use their posts to do the same or influence the politicians themselves.

Sensor-based analysis of human kinematics is valuable for providing personalized and detailed feedback when learning complex psychomotor skills. Kinematics examines the motion of a system of objects without directly considering the underlying forces. A visualization approach for a dynamic system such as complex human motion needs to account for changes over time.
In a learning scenario, highlighting differences between the performed move and a pedagogical framework is important for conveying necessary corrections to be undertaken. The goal of this thesis is to build a visualization system to show the movements and highlight potential errors on specific limbs/joints as meaningful feedback.