Page: 1 2 3 4 5 6 7 8 9... 71
81 |
Embedding Time Differences in Context-sensitive Neural Networks for Learning Time to Event ...
|
|
|
|
BASE
|
|
Show details
|
|
82 |
A Span-based Dynamic Local Attention Model for Sequential Sentence Classification ...
|
|
|
|
BASE
|
|
Show details
|
|
83 |
Learning from the Worst: Dynamically Generated Datasets to Improve Online Hate Detection ...
|
|
|
|
BASE
|
|
Show details
|
|
84 |
2A: Sentiment Analysis, Stylistic Analysis, and Argument Mining #1 ...
|
|
|
|
BASE
|
|
Show details
|
|
85 |
On Sample Based Explanation Methods for NLP: Faithfulness, Efficiency and Semantic Evaluation ...
|
|
|
|
BASE
|
|
Show details
|
|
86 |
How effective is BERT without word ordering? Implications for language understanding and data privacy ...
|
|
|
|
BASE
|
|
Show details
|
|
87 |
GEM: Natural Language Generation, Evaluation, and Metrics - Part 4 ...
|
|
|
|
BASE
|
|
Show details
|
|
88 |
The statistical advantage of automatic NLG metrics at the system level ...
|
|
|
|
Abstract:
Read paper: https://www.aclanthology.org/2021.acl-long.533 Abstract: Estimating the expected output quality of generation systems is central to NLG. This paper qualifies the notion that automatic metrics are not as good as humans in estimating system-level quality. Statistically, humans are unbiased, high variance estimators, while metrics are biased, low variance estimators. We compare these estimators by their error in pairwise prediction (which generation system is better?) using the bootstrap. Measuring this error is complicated: predictions are evaluated against noisy, human predicted labels instead of the ground truth, and metric predictions fluctuate based on the test sets they were calculated on. By applying a bias-variance-noise decomposition, we adjust this error to a noise-free, infinite test set setting. Our analysis compares the adjusted error of metrics to humans and a derived, perfect segment-level annotator, both of which are unbiased estimators dependent on the number of judgments collected. ...
|
|
Keyword:
Computational Linguistics; Condensed Matter Physics; Deep Learning; Electromagnetism; FOS Physical sciences; Information and Knowledge Engineering; Neural Network; Semantics
|
|
URL: https://underline.io/lecture/25762-the-statistical-advantage-of-automatic-nlg-metrics-at-the-system-level https://dx.doi.org/10.48448/zqan-b802
|
|
BASE
|
|
Hide details
|
|
90 |
Supporting Cognitive and Emotional Empathic Writing of Students ...
|
|
|
|
BASE
|
|
Show details
|
|
91 |
What's in the Box? An Analysis of Undesirable Content in the Common Crawl Corpus ...
|
|
|
|
BASE
|
|
Show details
|
|
92 |
Are Pretrained Convolutions Better than Pretrained Transformers? ...
|
|
|
|
BASE
|
|
Show details
|
|
93 |
Evaluation Examples are not Equally Informative: How should that change NLP Leaderboards? ...
|
|
|
|
BASE
|
|
Show details
|
|
94 |
Beyond Offline Mapping: Learning Cross-lingual Word Embeddings through Context Anchoring ...
|
|
|
|
BASE
|
|
Show details
|
|
95 |
Hate Speech Detection Based on Sentiment Knowledge Sharing ...
|
|
|
|
BASE
|
|
Show details
|
|
96 |
Taming Pre-trained Language Models with N-gram Representations for Low-Resource Domain Adaptation ...
|
|
|
|
BASE
|
|
Show details
|
|
97 |
Tail-to-Tail Non-Autoregressive Sequence Prediction for Chinese Grammatical Error Correction ...
|
|
|
|
BASE
|
|
Show details
|
|
98 |
WikiSum: Coherent Summarization Dataset for Efficient Human-Evaluation ...
|
|
|
|
BASE
|
|
Show details
|
|
99 |
An End-to-End Progressive Multi-Task Learning Framework for Medical Named Entity Recognition and Normalization ...
|
|
|
|
BASE
|
|
Show details
|
|
Page: 1 2 3 4 5 6 7 8 9... 71
|
|