AVOCADO: Visualization of Workflow–Derived Data Provenance for Reproducible Biomedical Research

185

146

MLModel ML Input ML Output Model State XAI Method XAI Input XAI Output explanations (visualizations, verbalizations, surrogate models, etc.)Abstract-We propose a framework for interactive and explainable machine learning that enables users to (1) understand machine learning models;(2) diagnose model limitations using different explainable AI methods; as well as (3) refine and optimize the models. Our framework combines an iterative XAI pipeline with eight global monitoring and steering mechanisms, including quality monitoring, provenance tracking, model comparison, and trust building. To operationalize the framework, we present explAIner, a visual analytics system for interactive and explainable machine learning that instantiates all phases of the suggested pipeline within the commonly used TensorBoard environment. We performed a user-study with nine participants across different expertise levels to examine their perception of our workflow and to collect suggestions to fill the gap between our system and framework. The evaluation confirms that our tightly integrated system leads to an informed machine learning process while disclosing opportunities for further extensions.

Section: Global Monitoring and Steering Mechanismsmentioning

confidence: 99%

explAIner: A Visual Analytics Framework for Interactive and Explainable Machine Learning

Spinner

Schlegel

Schäfer

et al. 2019

185

146

“…However, we also provide an expert-interaction mode that shows a wide range of controls for more advanced adaptations of the visualization, or of the algorithm. Lastly, to ensure data-provenance [64] tracking and trust-building, we track and show all user interactions, displaying all instances of optimization and direct model-manipulations in the timeline view (see Sect. 4.3).…”

Section: Visual Analytics Workpacementioning

confidence: 99%

Visual Analytics for Topic Model Optimization based on User-Steerable Speculative Execution

El‐Assady

Sperrle

Deußen

et al. 2019

To effectively assess the potential consequences of human interventions in model-driven analytics systems, we establish the concept of speculative execution as a visual analytics paradigm for creating user-steerable preview mechanisms. This paper presents an explainable, mixed-initiative topic modeling framework that integrates speculative execution into the algorithmic decisionmaking process. Our approach visualizes the model-space of our novel incremental hierarchical topic modeling algorithm, unveiling its inner-workings. We support the active incorporation of the user's domain knowledge in every step through explicit model manipulation interactions. In addition, users can initialize the model with expected topic seeds, the backbone priors. For a more targeted optimization, the modeling process automatically triggers a speculative execution of various optimization strategies, and requests feedback whenever the measured model quality deteriorates. Users compare the proposed optimizations to the current model state and preview their effect on the next model iterations, before applying one of them. This supervised human-in-the-loop process targets maximum improvement for minimum feedback and has proven to be effective in three independent studies that confirm topic model quality improvements.

“…Another exception in provenance retrieval is the well-studied field of workflow provenance graphs [41]. Due to workflow modifications or multiple executions with different parameters or input datasets, this type of provenance graph can rapidly grow very complex.…”

Section: Index and Provenancementioning

confidence: 99%

KnowledgePearls: Provenance-Based Visualization Retrieval

Stitz

Gratzl

Piringer

et al. 2019

Storing analytical provenance generates a knowledge base with a large potential for recalling previous results and guiding users in future analyses. However, without extensive manual creation of meta information and annotations by the users, search and retrieval of analysis states can become tedious. We present KnowledgePearls, a solution for efficient retrieval of analysis states that are structured as provenance graphs containing automatically recorded user interactions and visualizations. As a core component, we describe a visual interface for querying and exploring analysis states based on their similarity to a partial definition of a requested analysis state. Depending on the use case, this definition may be provided explicitly by the user by formulating a search query or inferred from given reference states. We explain our approach using the example of efficient retrieval of demographic analyses by Hans Rosling and discuss our implementation for a fast look-up of previous states. Our approach is independent of the underlying visualization framework. We discuss the applicability for visualizations which are based on the declarative grammar Vega and we use a Vega-based implementation of Gapminder as guiding example. We additionally present a biomedical case study to illustrate how KnowledgePearls facilitates the exploration process by recalling states from earlier analyses.