The use and creation of machine‐learning‐based solutions to solve problems or reduce their computational costs are becoming increasingly widespread in many domains. Deep Learning plays a large part in this growth. However, it has drawbacks such as a lack of explainability and behaving as a black‐box model. During the last few years, Visual Analytics has provided several proposals to cope with these drawbacks, supporting the emerging eXplainable Deep Learning field. This survey aims to (i) systematically report the contributions of Visual Analytics for eXplainable Deep Learning; (ii) spot gaps and challenges; (iii) serve as an anthology of visual analytical solutions ready to be exploited and put into operation by the Deep Learning community (architects, trainers and end users) and (iv) prove the degree of maturity, ease of integration and results for specific domains. The survey concludes by identifying future research challenges and bridging activities that are helpful to strengthen the role of Visual Analytics as effective support for eXplainable Deep Learning and to foster the adoption of Visual Analytics solutions in the eXplainable Deep Learning community. An interactive explorable version of this survey is available online athttps://aware‐diag‐sapienza.github.io/VA4XDL.

Point set visualization is required in lots of visualization techniques. Scatter plots as well as geographic heat-maps are straightforward examples. Data analysts are now well trained to use such visualization techniques. The availability of larger and larger datasets raises the need to make these techniques scale as fast as the data grows. The Big Data Infrastructure offers the possibility to scale horizontally. Designing point set visualization methods that fit into that new paradigm is thus a crucial challenge. In this paper, we present a complete architecture which fully fits into the Big Data paradigm and so enables interactive visualization of heatmaps at ultra-scale. A new distributed algorithm for multi-scale aggregation of point set is given and an adaptive GPU based method for kernel density estimation is proposed. A complete prototype working with Hadoop, HBase, Spark and WebGL has been implemented. We give a benchmark of our solution on a dataset having more than 2 billion points.

Visualization of graphs containing many nodes and edges efficiently is quite challenging since representations generally suffer from visual clutter induced by the large amount of edge crossings and node-edge overlaps. That problem becomes even more important when nodes positions are fixed, such as in geography were nodes positions are set according to geographical coordinates. Edge bundling techniques can help to solve this issue by visually merging edges along common routes but it can also help to reveal high-level edge patterns in the network and therefore to understand its overall organization. In this paper, we present a generalization of [18] to reduce the clutter in a 3D representation by routing edges into bundles as well as a GPU-based rendering method to emphasize bundles densities while preserving edge color. To visualize geographical networks in the context of the globe, we also provide a new technique allowing to bundle edges around and not across it.

The tools that are available to draw and to manipulate the representations of metabolism are usually restricted to metabolic pathways. This limitation becomes problematic when studying processes that span several pathways. The various attempts that have been made to draw genome-scale metabolic networks are confronted with two shortcomings: 1- they do not use contextual information which leads to dense, hard to interpret drawings, 2- they impose to fit to very constrained standards, which implies, in particular, duplicating nodes making topological analysis considerably more difficult. We propose a method, called MetaViz, which enables to draw a genome-scale metabolic network and that also takes into account its structuration into pathways. This method consists in two steps: a clustering step which addresses the pathway overlapping problem and a drawing step which consists in drawing the clustered graph and each cluster. The method we propose is original and addresses new drawing issues arising from the no-duplication constraint. We do not propose a single drawing but rather several alternative ways of presenting metabolism depending on the pathway on which one wishes to focus. We believe that this provides a valuable tool to explore the pathway structure of metabolism.