Try   HackMD

NLP xai

https://arxiv.org/pdf/2010.00711.pdf
總結目前NLP相關paper使用的解釋方法有哪些

  • Explanation derivation
    – 數學上的解釋
    Feature importances

  • Explanation presentation
    – 如何呈現這些Explanation
    Visualization
    for example: Saliency(顯著性，把相關句子或詞，標註顏色)

Local vs Global,
Self-Explaining(模型本身可解釋) vs Post-Hoc(預測後額外一些動作來解釋，舉例:LIME, surrogate model)

https://volpato.io/articles/1907-nlp-xai.html

http://people.csail.mit.edu/tommi/papers/AlvJaa_EMNLP2017.pdf

Lime for multi-classes classifier

https://marcotcr.github.io/lime/tutorials/Lime - multiclass.html

LimeTextExplainer

https://towardsdatascience.com/what-makes-your-question-insincere-in-quora-26ee7658b010

thesis

http://ceur-ws.org/Vol-2646/18-paper.pdf

Use Lime and SHAP to explain NLP models

https://towardsdatascience.com/explain-nlp-models-with-lime-shap-5c5a9f84d59b

Last changed by 
 
Explainable AI
0
315

Read more

Questioning the AI: Informing Design Practices for Explainable AI User Experiences

Dec 9, 2020
讀書會時程規劃

閱讀書籍 - Interpretable Machine Learning A Guide for Making Black Box Models Explainable. 兩篇論文： https://arxiv.org/pdf/2001.02478.pdf https://arxiv.org/pdf/1909.03012.pdf 書籍網址 https://christophm.github.io/interpretable-ml-book/ 倚任副理建議直接先閱讀第五章和第六章 讀書計畫初稿如下表，從下週二開始分享，之後視情況可機動調整

Nov 26, 2020
6.4 Influential Instances

機器學習的模型是訓練數據的產出，刪除任一訓練數據會影響訓練結果。若刪除每一訓練數據對模型產生巨大影響，則稱這個點唯有影響的點(instance)。對有影響的點分析可以幫助我們檢視模型。 Deletion Diagnostics : delete the instance from the training data, retrain the model on the reduced training dataset and observe the difference in the model parameters or predictions Influenced functions : upweight a data instance by approximating the parameter changes based on the gradients of the model parameters. 6.4.1 Deletion Diagnostics DFBETA : 衡量移除某個instance對模型參數的影響。 $DFBETA_i = β-β^{-i}$ 適用於有參數的模型，如 logistic regression or neural networks.

Nov 6, 2020
6.1 Counterfactual Explanations

A counterfactual explanation of a prediction describes the smallest change to the feature values that changes the prediction to a predefined output. How do we define a good counterfactual explanation? counterfactual instance produces the predefined prediction as closely as possible counterfactual should be as similar as possible to the instance regarding feature values a counterfactual instance should have feature values that are likely

Oct 20, 2020
Read more from Explainable AI
Published on HackMD