Rules:

1. Participants should prepare talks on the selected topic and present them during the meeting on the selected day (1 talk is required to pass with the satisfactory grade 3.0, 2 or more talks are required for higher grades).

2. The regular talk should take about 45 minutes; in special cases of some advanced topics, the talk may take 90 minutes (after prior arrangements with the instructor).

3. The talk should be accompanied by a presentation (slides); the presentation should be send to the instructor by email in 24h after the talk.

4. The presence is mandatory (max. 2 absences in the semester).

5. The final evaluation will be based on the talk, the presentation, the participation in the discussions (during talks of other participants) and the possible additional contribution.

Current Program of the Seminar:

GNN4TS. GNNs for Time Series with focus on Dynamic GNNs with application in RS (Klaudia Balcer, February 28, 2024) SLIDES

Introduction to Trajectory Data Mining (Kacper Puchalski, March 13, 2024)

Introduction to Self- and Semi-Supervised Learning (Maciej Malicki, March 20, 2024)

Self-supervised Graph Learning for Recommendation (Maurycy Borkowski, March 20, 2024)

Simple Graph Contrastive Learning for Recommendation (Arif Eftekhar Ahmed, March 27, 2024)

on Diffusion Models (Martyna Firgolska, March 27, 2024)

on deep learning for financial time series in ultra-high frequency (Michał Doros, April 3, 2024)

on UNets, their extensions and applications (Witold Płecha, April 10, 2024)

Dynamic Graph Representation via Self-Attention Network (Michał Zieliński, April 17, 2024)

Bayesian Graph Convolutional Neural Networks using Node Copying (Valery Tarasenko, April 17, 2024)

Suggestions for Next Talks (presenters welcome):

Introduction to Self- and Semi-Supervised Learning

Introduction to Representation Learning

Introduction to State Space Models

Introduction to Bayesian/Variational Learning (Variational Autoencoders, etc.)

Introduction to Trajectory Data Mining

... and see below for more advanced topics

Main Areas of Interests (the list will be updated during the semester):

Self- and Semi-Supervised Learning

- Semi-supervised Learning with Deep Generative Models [PDF]

- Self-supervised Learning: Generative or Contrastive [PDF]

- Semi-Supervised Learning with Ladder Networks [PDF]

- Debiased Self-Training for Semi-Supervised Learning [PDF]

Temporal Data Mining

- Discovering group dynamics in synchronous time series via hierarchical recurrent switching-state models [PDF]

- Spectral Temporal Graph Neural Network for Multivariate Time-series Forecasting [PDF]

Bayesian or Dynamic Graph Neural Networks

- A Survey of Graph Neural Networks for Recommender Systems: Challenges, Methods, and Directions (selected topics) [PDF]

- Gated Graph Sequence Neural Networks [PDF]

- Link Prediction Based on Graph Neural Networks [PDF]

- Bayesian Flow Networks [PDF]

- Bayesian Graph Neural Networks with Adaptive Connection Sampling [PDF]

- Bayesian Graph Convolutional Neural Networks using Node Copying [PDF]

- Variational Graph Recurrent Neural Networks [PDF]

- Graph Posterior Network: Bayesian Predictive Uncertainty for Node Classification [PDF]

- Dynamic Graph Representation Learning via Self-Attention Networks [PDF]

- Understanding Non-linearity in Graph Neural Networks from the Bayesian-Inference Perspective [PDF]

- Temporal Graph Neural Networks for Irregular Data [PDF]

- Temporal Graph Networks for Deep Learning on Dynamic Graphs [PDF]

Advanced Recommender Systems

- Leveraging Large Language Models for Sequential Recommendation [PDF]

- ProtoCF: Prototypical Collaborative Filtering for Few-shot Recommendation

Applications

- Spatio-Temporal Graph Transformer Networks for Pedestrian Trajectory Prediction [PDF]

- Evolution of Image Segmentation using Deep Convolutional Neural Network: A Survey [PDF]

- Semi-supervised Change Detection of Small Water Bodies Using RGB and Multispectral Images in Peruvian Rainforests [PDF]

- Flood Segmentation on Sentinel-1 SAR Imagery with Semi-Supervised Learning [PDF]

- Productive Crop Field Detection: A New Dataset and Deep Learning Benchmark Results [PDF]

- Comparative performance analysis of simple U-Net, residual attention U-Net, and VGG16-U-Net for inventory inland water bodies [PDF]