iNNspector: Visual, Interactive Deep Model Debugging

TL;DR

iNNspector is a visual, interactive tool that helps deep learning developers systematically analyze and debug models by exploring internal data across multiple abstraction levels, improving understanding and refinement.

Contribution

The paper introduces a structured data framework for deep learning experiments and the iNNspector system, enabling interactive visualization and analysis for debugging.

Findings

Effective visualization of model internals across abstraction levels.

Improved debugging efficiency demonstrated in real-world use-cases.

Positive user feedback on usability and utility.

Abstract

Deep learning model design, development, and debugging is a process driven by best practices, guidelines, trial-and-error, and the personal experiences of model developers. At multiple stages of this process, performance and internal model data can be logged and made available. However, due to the sheer complexity and scale of this data and process, model developers often resort to evaluating their model performance based on abstract metrics like accuracy and loss. We argue that a structured analysis of data along the model's architecture and at multiple abstraction levels can considerably streamline the debugging process. Such a systematic analysis can further connect the developer's design choices to their impacts on the model behavior, facilitating the understanding, diagnosis, and refinement of deep learning models. Hence, in this paper, we (1) contribute a conceptual framework structuring the data space of deep learning experiments. Our framework, grounded in literature analysis and requirements interviews, captures design dimensions and proposes mechanisms to make this data explorable and tractable. To operationalize our framework in a ready-to-use application, we (2) present the iNNspector system. iNNspector enables tracking of deep learning experiments and provides interactive visualizations of the data on all levels of abstraction from multiple models to individual neurons. Finally, we (3) evaluate our approach with three real-world use-cases and a user study with deep learning developers and data analysts, proving its effectiveness and usability.