• Corpus ID: 462623

Mind the Gap: A Generative Approach to Interpretable Feature Selection and Extraction

@inproceedings{Kim2015MindTG,
  title={Mind the Gap: A Generative Approach to Interpretable Feature Selection and Extraction},
  author={Been Kim and Julie A. Shah and Finale Doshi-Velez},
  booktitle={NIPS},
  year={2015}
}
We present the Mind the Gap Model (MGM), an approach for interpretable feature extraction and selection. By placing interpretability criteria directly into the model, we allow for the model to both optimize parameters related to interpretability and to directly report a global set of distinguishable dimensions to assist with further data exploration and hypothesis generation. MGM extracts distinguishing features on real-world datasets of animal features, recipes ingredients, and disease co… 

Figures and Tables from this paper

The Need for Interpretable Features: Motivation and Taxonomy

It is claimed that the term “interpretable feature” is not specific nor detailed enough to capture the full extent to which features impact the usefulness of ML explanations, and a formal taxonomy is needed of the feature properties that may be required by domain experts taking real-world actions.

The Need for Interpretable Features

It is claimed that the term "interpretable feature" is not specific nor detailed enough to capture the full extent to which features impact the usefulness of ML explanations and a formal taxonomy is needed of the feature properties that may be required by domain experts taking real-world actions.

Inducing Semantic Grouping of Latent Concepts for Explanations: An Ante-Hoc Approach

This work worked with one of such models motivated by explicitly representing the classifier function as a linear function and showed that by exploiting probabilistic latent and properly modifying different parts of the model can result better explanation as well as provide superior predictive performance.

Interpretability Beyond Feature Attribution: Quantitative Testing with Concept Activation Vectors (TCAV)

Concept Activation Vectors (CAVs) are introduced, which provide an interpretation of a neural net's internal state in terms of human-friendly concepts, and may be used to explore hypotheses and generate insights for a standard image classification network as well as a medical application.

Towards Ground Truth Explainability on Tabular Data

Copulas, a concise specification of the desired statistical properties of a dataset, so that users can build intuition around explainability using controlled data sets and experimentation, is proposed for tabular data.

Multi-Dimensional Explanation of Target Variables from Documents

The Multi-Target Masker (MTM) is proposed, a soft multi-dimensional mask that models a relevance probability distribution over the set of target variables to handle ambiguities and is the first to also achieve the highest F1 scores for all the target variables simultaneously.

Considerations for Evaluation and Generalization in Interpretable Machine Learning

A definitions of interpretability are discussed and described and when interpretability is needed (and when it is not) and a taxonomy for rigorous evaluation is described, and recommendations for researchers are made.

Rationalization through Concepts

Experiments show that ConRAT is the first to generate concepts that align with human rationalization while using only the overall label, and it outperforms state-of-the-art methods trained on each aspect label independently.

Interacting with Predictions: Visual Inspection of Black-box Machine Learning Models

The design and implementation of an interactive visual analytics system, Prospector, that provides interactive partial dependence diagnostics and support for localized inspection allows data scientists to understand how and why specific datapoints are predicted as they are.

Explainable Artificial Intelligence: a Systematic Review

This systematic review contributes to the body of knowledge by clustering these methods with a hierarchical classification system with four main clusters: review articles, theories and notions, methods and their evaluation.
...

References

SHOWING 1-10 OF 33 REFERENCES

The Bayesian Case Model: A Generative Approach for Case-Based Reasoning and Prototype Classification

The Bayesian Case Model is presented, a general framework for Bayesian case-based reasoning (CBR) and prototype classification and clustering and subspace representation that provides quantitative benefits in interpretability while preserving classification accuracy.

Contrastive Learning Using Spectral Methods

This paper formalizes this notion of contrastive learning for mixture models, and develops spectral algorithms for inferring mixture components specific to a foreground data set when contrasted with a background data set.

Comprehensible classification models: a position paper

This paper discusses the interpretability of five types of classification models, namely decision trees, classification rules, decision tables, nearest neighbors and Bayesian network classifiers, and the drawbacks of using model size as the only criterion to evaluate the comprehensibility of a model.

Learning Determinantal Point Processes

This thesis shows how determinantal point processes can be used as probabilistic models for binary structured problems characterized by global, negative interactions, and demonstrates experimentally that the techniques introduced allow DPPs to be used for real-world tasks like document summarization, multiple human pose estimation, search diversification, and the threading of large document collections.

Clustering with the Fisher Score

This paper develops a novel but simple clustering algorithm specialized for the Fisher score, which can exploit important dimensions and is successfully tested in experiments with artificial data and real data.

Fully Sparse Topic Models

This paper shows that FSTM can perform substantially better than various existing topic models by different performance measures, and provides a principled way to directly trade off sparsity of solutions against inference quality and running time.

Sparse Additive Generative Models of Text

This approach has two key advantages: it can enforce sparsity to prevent overfitting, and it can combine generative facets through simple addition in log space, avoiding the need for latent switching variables.

Unsupervised Variable Selection: when random rankings sound as irrelevancy

This paper proposes to combine multiple ranking to go ahead toward a stable consensus variable subset in a totally unsupervised fashion.

Priors for Diversity in Generative Latent Variable Models

This work revisits independence assumptions for probabilistic latent variable models with a determinantal point process (DPP), leading to better intuition for the latent variable representation and quantitatively improved unsupervised feature extraction, without compromising the generative aspects of the model.

The discovery of structural form

This work presents a computational model that learns structures of many different forms and that discovers which form is best for a given dataset and brings structure learning methods closer to human abilities and may lead to a deeper computational understanding of cognitive development.