Published Aug 13, 2020

117 - Interpreting NLP Model Predictions, with Sameer Singh

Sameer Singh delves into the complexities of interpreting NLP model predictions, discussing the challenges of nonlinearity, perturbation techniques, and the critical need for effective evaluation methods to enhance model transparency and trust. He explores feature attribution, influence functions, and explanation generation, shedding light on the evolving advancements making model outputs more understandable and reliable.

Episode Highlights

Topics covered

Episode Highlights

Related Episodes

114 - Behavioral Testing of NLP Models, with Marco Tulio Ribeiro
Answers 383 questions
87 - Pathologies of Neural Models Make Interpretation Difficult, with Shi Feng
Answers 383 questions
109 - What Does Your Model Know About Language, with Ellie Pavlick
Answers 383 questions
130 - Linking human cognitive patterns to NLP Models, with Lisa Beinborn
Answers 383 questions
91 - (Executable) Semantic Parsing, with Jonathan Berant
Answers 383 questions
119 - Social NLP, with Diyi Yang
Answers 383 questions
106 - Ethical Considerations In NLP Research, with Emily Bender
Answers 383 questions
93 - NLP/ML for clinical data, with Alistair Johnson
Answers 383 questions
53 - Classical Structured Prediction Losses for Sequence to Sequence Learning, with Sergey and Myle
Answers 383 questions
81 - BlackboxNLP, with Afra Alishahi and Tal Linzen
Answers 383 questions
124 - Semantic Machines and Task-Oriented Dialog, with Jayant Krishnamurthy and Hao Fang
Answers 383 questions
33 - Entity Linking via Joint Encoding of Types, Descriptions, and Context, with Nitish Gupta
Answers 383 questions
37 - On Statistical Significance, Training Variance, and Why Reporting Score Distributions Matters
Answers 383 questions
07 - Capturing Semantic Similarity for Entity Linking with Convolutional Neural Networks
Answers 383 questions
44 - Truly Low Resource NLP, with Anders Søgaard
Answers 383 questions

Dexa/NLP Highlights

117 - Interpreting NLP Model Predictions, with Sameer Singh

Topics covered

Popular Clips

Model Interpretability Challenges

Interpreting Complex Models

Explanation Generation Methods

Influence Functions

Feature Attribution Methods

Model Influence Analysis

Research Opportunities

NLP Challenges

Exploring NLP Interpretability

Influence Functions in NLP

Explanation Techniques

Model Evaluation Insights

Understanding Perturbation Methods

Perturbation Explanations

Episode Highlights

Challenges in Interpretability

Evaluating Explanations

Interpretability Methods

Related Episodes

114 - Behavioral Testing of NLP Models, with Marco Tulio Ribeiro

87 - Pathologies of Neural Models Make Interpretation Difficult, with Shi Feng

109 - What Does Your Model Know About Language, with Ellie Pavlick

130 - Linking human cognitive patterns to NLP Models, with Lisa Beinborn

91 - (Executable) Semantic Parsing, with Jonathan Berant

119 - Social NLP, with Diyi Yang

106 - Ethical Considerations In NLP Research, with Emily Bender

93 - NLP/ML for clinical data, with Alistair Johnson

53 - Classical Structured Prediction Losses for Sequence to Sequence Learning, with Sergey and Myle

81 - BlackboxNLP, with Afra Alishahi and Tal Linzen

124 - Semantic Machines and Task-Oriented Dialog, with Jayant Krishnamurthy and Hao Fang

33 - Entity Linking via Joint Encoding of Types, Descriptions, and Context, with Nitish Gupta

37 - On Statistical Significance, Training Variance, and Why Reporting Score Distributions Matters

07 - Capturing Semantic Similarity for Entity Linking with Convolutional Neural Networks

44 - Truly Low Resource NLP, with Anders Søgaard

117 - Interpreting NLP Model Predictions, with Sameer Singh

Topics covered

Popular Clips

Episode Highlights

Challenges in InterpretabilitySameer Singh explores the complexities of interpreting NLP model predictions, focusing on challenges posed by nonlinearity and perturbation techniques. He highlights the difficulties in defining explanations and ensuring perturbation methods produce valid instances.

Challenges in Interpretability

Evaluating ExplanationsSameer Singh, an Assistant Professor of Computer Science at UC Irvine, explores the intricacies of interpreting NLP model predictions. He discusses the importance of evaluation techniques and user-centric assessments in enhancing model transparency and trust.

Evaluating Explanations

Interpretability MethodsSameer Singh explores the intricacies of interpreting NLP model predictions, focusing on feature attribution, influence functions, and explanation generation. He highlights the challenges and advancements in making model outputs understandable and reliable.

Interpretability Methods

Related Episodes