Seeing is Learning in High Dimensions |Alexandru Telea Keynote Machine Learning & Data Visualization

Dimensionality Reduction Bridges Machine Learning and Data Visualization In this insightful keynote presentation, Professor Alexandru Telea explores the fascinating intersection of machine learning, high-dimensional data visualization, and dimensionality reduction techniques, revealing how visual analytics can help us better understand complex AI models and how machine learning can, in turn, improve visualization methods. Modern computer vision systems, machine learning models, and advanced sensing technologies generate enormous volumes of high-dimensional data. Understanding these datasets remains one of the biggest challenges in AI and data science. In this talk, Professor Telea demonstrates how multidimensional projection techniques such as PCA, t-SNE, UMAP, and other dimensionality reduction methods can serve as powerful tools for exploring, interpreting, and explaining machine learning models. Key Topics Covered • Understanding high-dimensional data in machine learning and computer vision • The role of dimensionality reduction in visual data exploration • Using projections to interpret neural network behavior and hidden representations • Visualizing feature spaces, class separation, and model training dynamics • Identifying hidden biases and unintended learning patterns in neural networks • Decision boundary visualization and explainable AI (XAI) techniques • Measuring and benchmarking projection quality across dozens of dimensionality reduction algorithms • Leveraging machine learning to create faster and more scalable projection methods • Neural-network-based projection techniques capable of handling millions of data points • Future directions in visual analytics, explainable machine learning, and high-dimensional visualization Throughout the keynote, Professor Telea presents practical examples showing how visualization can uncover hidden structures inside machine learning models, improve classifier performance, reveal model weaknesses, and provide valuable insights that traditional performance metrics alone cannot capture. About the Speaker Professor Alexandru Telea is a leading researcher in visual analytics and data visualization and serves as Professor of Visual Data Analytics in the Department of Information and Computing Sciences at Utrecht University. With more than 25 years of experience in visualization research, he has played major leadership roles in international conferences including EuroVis, VISSOFT, SoftVis, IVAPP, and SIBGRAPI. His research focuses on information visualization, scientific visualization, high-dimensional data analysis, and visual analytics for machine learning. He is also the author of the widely recognized book Data Visualization: Principles and Practice. Why Watch This Talk? Whether you are a researcher, data scientist, machine learning engineer, AI practitioner, visualization expert, or graduate student, this keynote provides valuable insights into how visual exploration and machine learning can work together to improve model understanding, interpretability, and performance. #MachineLearning #DataVisualization #DimensionalityReduction #ExplainableAI #XAI #ComputerVision #HighDimensionalData #VisualAnalytics #DeepLearning #NeuralNetworks #tSNE #UMAP #PCA #ArtificialIntelligence #DataScience #AlexandruTelea #VisualizationResearch #AIInterpretability #MLResearch #ConferenceKeynote