The recent surge of machine learning (ML) has opened up various opportunities when analyzing big datasets. Beyond basic, non-ML supported techniques of big data analytics, such as identifying similar items in big datasets, or arranging how to distribute jobs on large compute clusters, for example, the ML supported techniques enable to extract knowledge from large datasets at utmost diversity and accuracy.

The seminar will start with a mini lecture. First, lectures will explain how to cluster datasets. Clustering is an 'unsupervised' machine learning technique by which to mine social network graphs, for example. Second, 'supervised' machine learning techniques (where 'deep learning' likely is the most prominent recent technique) and their use in analyzing big data will be discussed. The mini lecture will be followed by seminar presentations, to be presented in small groups of 1-2 students.

• Lernraum-Plus
• Seminar Recordings

You can selected a paper you wish to present In the Lernraum-Plus

Meant to serve as a reference for the students to familiarize themselves with

• Graph neural nets: A review of methods and applications, Zhou et. al
• A Comprehensive Survey on Graph Neural Nets, Wu et. al
• Graph Signal Processing: Overview, Challenges and Applications, Ortega et. al
• Graph Representation Learning in Biomedicine, Li et. al
• Graph Representation Learning for Single Cell Biology, Hetzel et. al

Papers that describe different architectures, etc. These can be presented.

• Graph Attention Networks, Veličković et. al
• Variational Graph Autoencoders, Kipf, T. and Welling, M.
• Semi-Supervised Classification with Graph Convolutional Network, Kipf, T. and Welling, M.
• Graph Normalizing Flows, Liu et. al
• MagNet: A Neural Network for Directed Graphs, Zhang et. al
• Residual Gated Graph Convnets, Bresson, X. and Laurent, T.
• DIGRAC: Digraph Clustering Based on Flow Imbalance, He et. al

These papers can be presented too.

• Graph Convolutional Matrix Completion, van den Berg et. al
• Hyper-SAGNN: a self-attention based graph neural network for hypergraphs, Zhang et. al
• Learning to Untangle Genome Assembly with Graph Convolutional Networks, Vrček et. al
• Directional Message Passing for Molecular Graphs, Gasteiger et. al
• ETA Prediction with Graph Neural Networks in Google Maps, Derrow-Pinion et. al
• An Efficient Graph Convolutional Network Technique for the Travelling Salesman Problem, Joshi et. al
• eFraudCom: An E-commerce Fraud Detection System via Competitive Graph Neural Networks, Zhang et. al