• Corpus ID: 3911355

Explanation and Justification in Machine Learning : A Survey Or

  title={Explanation and Justification in Machine Learning : A Survey Or},
  author={Or Biran and Courtenay V. Cotton},
We present a survey of the research concerning explanation and justification in the Machine Learning literature and several adjacent fields. Within Machine Learning, we differentiate between two main branches of current research: interpretable models, and prediction interpretation and justification. 

Combinatorial Methods for Explainable AI

This short paper introduces an approach to producing explanations or justifications of decisions made by artificial intelligence and machine learning (AI/ML) systems, using methods derived from fault

Building More Explainable Artificial Intelligence With Argumentation

This paper proposes an argumentation-based approach to explainable AI, which has the potential to generate more comprehensive explanations than existing approaches.

Towards a Framework for Challenging ML-Based Decisions

The differences between explanations and justifications are highlighted and a framework to generate evidence to support or to dismiss challenges is outlined to challenge the results of an algorithmic decision system relying on machine learning.

Weight of Evidence as a Basis for Human-Oriented Explanations

This work takes a step towards reconciling machine explanations with those that humans produce and prefer by taking inspiration from the study of explanation in philosophy, cognitive science, and the social sciences.

A Survey on Explainability in Machine Reading Comprehension

This paper presents a systematic review of benchmarks and approaches for explainability in Machine Reading Comprehension (MRC), and presents the evaluation methodologies to assess the performance of explainable systems.

Towards making NLG a voice for interpretable Machine Learning

It is shown that self-reported rating of NLG explanation was higher than that for a non-NLG explanation, but when tested for comprehension, the results were not as clear-cut showing the need for performing more studies to uncover the factors responsible for high-quality NLG explanations.

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.

From Shallow to Deep Interactions Between Knowledge Representation, Reasoning and Machine Learning (Kay R. Amel group)

This paper proposes a tentative and original survey of meeting points between Knowledge Representation and Reasoning (KRR) and Machine Learning (ML), two areas which have been developing quite

Exploiting Language Instructions for Interpretable and Compositional Reinforcement Learning

This work attempts to interpret the latent space from an RL agent to identify its current objective in a complex language instruction and shows that the classification process causes changes in the hidden states which makes them more easily interpretable, but also causes a shift in zero-shot performance to novel instructions.

Natural Language Generation Challenges for Explainable AI

  • E. Reiter
  • Computer Science
    Proceedings of the 1st Workshop on Interactive Natural Language Technology for Explainable Artificial Intelligence (NL4XAI 2019)
  • 2019
This paper discusses the challenges of good quality explanations of artificial intelligence reasoning from a Natural Language Generation (NLG) perspective, and highlights four specific NLG for XAI research challenges.



Varieties of Justification in Machine Learning

Forms of justification for inductive machine learning techniques are discussed and classified into four types. This is done with a view to introduce some of these techniques and their justificatory

A review of explanation methods for Bayesian networks

The basic properties that characterise explanation methods are described and the methods developed to date for explanation in Bayesian networks are reviewed.

Human-Centric Justification of Machine Learning Predictions

This work proposes a novel approach to producing justifications that is geared towards users without machine learning expertise, focusing on domain knowledge and on human reasoning, and utilizing natural language generation.

Learning theory analysis for association rules and sequential event prediction

A theoretical analysis for prediction algorithms based on association rules, which introduces a problem for which rules are particularly natural, called "sequential...

"Why Should I Trust You?": Explaining the Predictions of Any Classifier

LIME is proposed, a novel explanation technique that explains the predictions of any classifier in an interpretable and faithful manner, by learning aninterpretable model locally varound the prediction.

Deriving Explanations and Implications for Constraint Satisfaction Problems

It is shown that consistency methods can be used to generate inferences that support both functions and explanations take the form of trees that show the basis for assignments and deletions in terms of previous selections.

Gaining insight through case-based explanation

A knowledge-light approach to case-based explanations that works by selecting cases based on explanation utility and offering insights into the effects of feature-value differences is examined.

Explaining Classifications For Individual Instances

It is demonstrated that the generated explanations closely follow the learned models and a visualization technique is presented that shows the utility of the approach and enables the comparison of different prediction methods.

Rationalizing Neural Predictions

The approach combines two modular components, generator and encoder, which are trained to operate well together and specifies a distribution over text fragments as candidate rationales and these are passed through the encoder for prediction.

Defining Explanation in Probabilistic Systems

An approach to defining a notion of "better explanation" is proposed that combines some of the features of both Gardenfors and Pearl together with more recent work by Pearl and others on causality.