Do Neural Networks Preserve Case Structure? Case-Based Decomposition, Interpretation, and Decision Consistency
Not on the map. Why?
Abstract (arXiv)
Neural networks increasingly inform consequential decisions, making their reliability increasingly important. Yet their internal mechanisms provide little evidence of whether decisions remain grounded in the training cases and which cases ultimately support or oppose their outcomes. Without this connection between decisions and training cases, users cannot determine whether a model has learned reliable decision patterns from data. This motivates a fundamental question: do neural networks preserve case structure? We establish a connection between neural networks and Case-Based Decision Theory (CBDT), showing that trained neural networks can preserve a recoverable case structure through their learned representations. Such a structure allows fitted decision margins to be decomposed into individual case contributions. We identify the conditions under which this recovered case structure admits a CBDT interpretation. We further establish decision consistency between this interpretation and the decision selected by the original neural network. Experiments on a controlled CBDT setting and three decision tasks based on real-world data validate our approach. These results connect neural network decisions with the cases that shape them. This connection allows model choices to be traced back to supporting and opposing cases, providing a basis for assessing the reliability of neural network decisions.
Key facts. Cited by 1 paper in the map (data through 7 Oct 2026). Builds on 25 of its 38 references that are in the map. First posted on arXiv 13 Jul 2026.
Builds on
25 of 38 references.
References · 25
- 2017A Unified Approach to Interpreting Model Predictionscited by 6,488
- 2016"Why Should I Trust You?": Explaining the Predictions of Any Classifiercited by 5,040
- 2017Axiomatic Attribution for Deep Networkscited by 3,350
- 2017Understanding Black-box Predictions via Influence Functionscited by 1,714
- 2023Sparse Autoencoders Find Highly Interpretable Features in Language Modelscited by 1,359
- 2018Sanity Checks for Saliency Mapscited by 1,149
- 2020Concept Bottleneck Modelscited by 1,022
- 2018This Looks Like That: Deep Learning for Interpretable Image Recognitioncited by 731
- 2020Estimating Training Data Influence by Tracing Gradient Descentcited by 565
- 2024Sparse Feature Circuits: Discovering and Editing Interpretable Causal Graphs in Language Modelscited by 412
- 2015Convergent Learning: Do different neural networks learn the same representations?cited by 215
- 2018Representer Point Selection for Explaining Deep Neural Networkscited by 187
13 more
- 2023Sudden Drops in the Loss: Syntax Acquisition, Phase Transitions, and Simplicity Bias in MLMscited by 121
- 2023DataInf: Efficiently Estimating Data Influence in LoRA-tuned LLMs and Diffusion Modelscited by 121
- 2023How Do Transformers Learn In-Context Beyond Simple Functions? A Case Study on Learning with Representationscited by 72
- 2022ProtoPFormer: Concentrating on Prototypical Parts in Vision Transformers for Interpretable Image Recognitioncited by 63
- 2023Intriguing Properties of Data Attribution on Diffusion Modelscited by 47
- 2024Scalable Influence and Fact Tracing for Large Language Model Pretrainingcited by 40
- 2024Capturing the Temporal Dependence of Training Data Influencecited by 34
- 2024Energy-Based Concept Bottleneck Models: Unifying Prediction, Concept Intervention, and Probabilistic Interpretationscited by 33
- 2024Efficient Sketches for Training Data Attribution and Studying the Loss Landscapecited by 10
- 2023Faithful and Efficient Explanations for Neural Networks via Neural Tangent Kernel Surrogate Modelscited by 8
- 2025Influence Dynamics and Stagewise Data Attributioncited by 5
- 2024Final-Model-Only Data Attribution with a Unifying View of Gradient-Based Methodscited by 4
- 2025Learning to Weight Parameters for Training Data Attributioncited by 3
Built on by
Citing papers per year: 2026: 1.
1 paper.