POEM: Pattern-Oriented Explanations of Convolutional Neural Networks

Summary: POEM: modular framework extracting semantic-concept patterns (shapes, colors) to explain CNN image classifiers. Yields interpretable rules like "if sofa then living room" linking concept presence + model attention to labels, with quantitative and qualitative advantages over prior explainers. (summarized by gpt-5-mini on Feb 09 2026)

Paper ID: 13156
Venue: VLDB
Year: 2023
Pagerank: 4.8925595e-05
Overall Rank: 6,903 | 51.98%
DOI: 10.14778/3611479.3611518

Incoming Non-self Citations Over Time

Authors

Incoming Citations (Sorted by Pagerank)

Showing 1 of 1 citing papers.

Rank	Citing Paper	Year	Venue	Pagerank
11,121	Demonstration of VCR: A Tabular Data Slicing Approach to Understanding Object Detection Model Performance	2024	VLDB	4.1945683e-05

Outgoing Citations (Sorted by Pagerank)

Showing 3 of 3 cited papers.

Citations counted here include only citations to other VLDB/SIGMOD/CIDR/PODS papers in this database.

Rank	Cited Paper	Year	Venue	Pagerank
2,863	Incremental and Approximate Inference for Faster Occlusion-based Deep CNN Explanations	2019	SIGMOD	7.9877991e-05
4,872	Explainable AI: Foundations, Applications, Opportunities for Data Management Research	2022	SIGMOD	5.8609352e-05
13,221	POEM: Pattern-Oriented Explanations of CNN Models	2022	VLDB	-

Semantically Similar Papers

Overall Rank	Paper	Year	Venue	Pagerank
3,293	Jointly Optimizing Preprocessing and Inference for DNN-based Visual Analytics	2021	VLDB	7.2629834e-05
11,482	Pool of Experts: Realtime Querying Specialized Knowledge in Massive Neural Networks	2021	SIGMOD	4.1945683e-05
3,206	Panorama: A Data System for Unbounded Vocabulary Querying over Video	2020	VLDB	7.3826363e-05
2,886	VISTA: Optimized System for Declarative Feature Transfer from Deep CNNs at Scale	2020	SIGMOD	7.9612767e-05
13,313	Demonstration of Krypton: Optimized CNN Inference for Occlusion-based Deep CNN Explanations	2019	VLDB	-
9,331	dCAM: Dimension-wise Class Activation Map for Explaining Multivariate Data Series Classification	2022	SIGMOD	4.3556432e-05
9,764	View-based Explanations for Graph Neural Networks	2024	SIGMOD	4.2856106e-05
11,170	Unsupervised Hashing with Semantic Concept Mining	2023	SIGMOD	4.1945683e-05
2,863	Incremental and Approximate Inference for Faster Occlusion-based Deep CNN Explanations	2019	SIGMOD	7.9877991e-05
13,221	POEM: Pattern-Oriented Explanations of CNN Models	2022	VLDB	-