In recent years, a wide spectrum of deep learning visualization methods has emerged. For a comprehensive overview of visual analytics (VA) for deep learning, we refer the reader to a survey by Hohman et al. A case study using a VA system to assess a model's performance to detect and classify traffic lights has shown that interactive VA systems can successfully guide experts to improve their training data While these examples all focus on the inspection of a single model, others support model comparisons. For example, using REMAP [cashman_ablate_2020], users can rapidly create model architectures through ablations (i.e., removing single layers of an existing model) and variations (i.e., creating new models through layer replacements) and compare the created models by their structure and performance. Interactive "playgrounds" require relatively little underlying deep learning knowledge and can be used for educational purposes.

While convolutional neural networks (CNNs) have found wide adoption as state-of-the-art models for image-related tasks, their predictions are often highly sensitive to small input perturbations, which the human vision is robust against. This paper presents Perturber, a web-based application that allows users to instantaneously explore how CNN activations and predictions evolve when a 3D input scene is interactively perturbed. Perturber offers a large variety of scene modifications, such as camera controls, lighting and shading effects, background modifications, object morphing, as well as adversarial attacks, to facilitate the discovery of potential vulnerabilities. Fine-tuned model versions can be directly compared for qualitative evaluation of their robustness. Case studies with machine learning experts have shown that Perturber helps users to quickly generate hypotheses about model vulnerabilities and to qualitatively compare model behavior. Using quantitative analyses, we could replicate users' insights with other CNN architectures and input images, yielding new insights about the vulnerability of adversarially trained models.

Embeddings of words and concepts capture syntactic and semantic regularities of language; however, they have seen limited use as tools to study characteristics of different corpora and how they relate to one another. We introduce TextEssence, an interactive system designed to enable comparative analysis of corpora using embeddings. TextEssence includes visual, neighbor-based, and similarity-based modes of embedding analysis in a lightweight, web-based interface. We further propose a new measure of embedding confidence based on nearest neighborhood overlap, to assist in identifying high-quality embeddings for corpus analysis. A case study on COVID-19 scientific literature illustrates the utility of the system. TextEssence is available from https://github.com/drgriffis/text-essence.