Towards Efficient Visual Simplification of Computational Graphs in Deep Neural Networks
This addresses the challenge for researchers and practitioners in efficiently understanding and debugging large DNN models, though it is incremental as it builds on existing visualization toolkits.
The paper tackles the problem of visualizing complex, large-scale computational graphs in deep neural networks (e.g., BERT) by proposing visual simplification techniques, resulting in a tool that reduces elements by 60% on average and improves model recognition and diagnosis.
A computational graph in a deep neural network (DNN) denotes a specific data flow diagram (DFD) composed of many tensors and operators. Existing toolkits for visualizing computational graphs are not applicable when the structure is highly complicated and large-scale (e.g., BERT [1]). To address this problem, we propose leveraging a suite of visual simplification techniques, including a cycle-removing method, a module-based edge-pruning algorithm, and an isomorphic subgraph stacking strategy. We design and implement an interactive visualization system that is suitable for computational graphs with up to 10 thousand elements. Experimental results and usage scenarios demonstrate that our tool reduces 60% elements on average and hence enhances the performance for recognizing and diagnosing DNN models. Our contributions are integrated into an open-source DNN visualization toolkit, namely, MindInsight [2].