ExplorerTree: A Focus+Context Exploration Approach for 2D Embeddings

In exploratory tasks involving high-dimensional datasets, dimensionality reduction (DR) techniques help analysts to discover patterns and other useful information. Although scatter plot representations of DR results allow for cluster identification similarity analysis, such a visual metaphor presents problems when the number instances dataset increases, resulting in cluttered visualizations. this work, we propose plot-based multilevel approach display address clutter-related visualizing large together with definition methodology use focus+context interaction on non-hierarchical embeddings. The proposed technique, called ExplorerTree, uses sampling selection technique plots reduce clutter guide users through tasks. We demonstrate ExplorerTree's effectiveness case, where visually explore activation images convolutional layers neural network. Finally, also conducted user experiment evaluate ability convey embedding structures using different strategies.