Incorporating Uncertainty into Deep Learning for Spoken Language Assessment

By Andrey Malinin, Anton Ragni, K. Knill, Mark John Francis Gales. 2017

47 citations recorded by the source corpus.

There is a growing demand for automatic assessment of spoken English proficiency. These systems need to handle large variations in input data owing to the wide range of candidate skill levels and L1s, and errors from ASR. Some candidates will be a poor match to the training data set, undermining the validity of the predicted grade. For high stakes tests it is essential for such systems not only to grade well, but also to provide a measure of their uncertainty in their predictions, enabling rejection to human graders. Previous work examined Gaussian Process (GP) graders which, though successful, do not scale well with large data sets. Deep Neural Network (DNN) may also be used to provide uncertainty using Monte-Carlo Dropout (MCD). This paper proposes a novel method to yield uncertainty and compares it to GPs and DNNs with MCD. The proposed approach explicitly teaches a DNN to have low uncertainty on training data and high uncertainty on generated artificial data. On experiments conducted on data from the Business Language Testing Service (BULATS), the proposed approach is found to outperform GPs and DNNs with MCD in uncertainty-based rejection whilst achieving comparable grading performance.

Subfields: Artificial Intelligence, Control and Systems Engineering

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Artificial Intelligence

1. Kimin Lee
2. Jinwoo Shin
3. Yixuan Li
4. Andrey Malinin
5. Mark John Francis Gales
6. Kibok Lee
7. Honglak Lee
8. Mohamed H. Zaki

In Control and Systems Engineering

1. Andrey Malinin
2. Mark John Francis Gales
3. Jinwoo Shin
4. Honglak Lee
5. Kimin Lee
6. Sergey Levine

Citations in the corpus

• Predictive Uncertainty Estimation via Prior Networks (2018, methodology): “One solution is to synthetically generate points on the boundary of the in-domain region using a generative model [21, 22].”
• Training Confidence-calibrated Classifiers for Detecting Out-of-Distribution Samples (2017, methodology): “n in our experiments, our method can be used for any classification tasks using deep neural networks. It is also interesting future directions applying our methods for other related tasks: regression (Malinin et al., 2017), network calibration (Guo et al., 2017), Bayesian probabilistic models (Li &am”
• High-Quality Prediction Intervals for Deep Learning: A Distribution-Free, Ensembled Approach (2018, background): “Another MVE extention encourages high uncertainty in data regions not observed by augmenting training data with synthetic ‘out-of-distribution’ samples of high variance (Malinin et al., 2017).”
• Uncertainty Estimation in Autoregressive Structured Prediction (2021, methodology): “1 PREDICTION REJECTION RATIO Here we describe the Prediction Rejection Ratio metric, proposed in (Malinin, 2019; Malinin et al., 2017), which in this work is used to assess how well measures of sequence-level uncertainty are able to identify sentences which are hard to translate/transcribe.”