Noise Contrastive Estimation and Negative Sampling for Conditional Models: Consistency and Statistical Efficiency
arXiv:1809.01812
Abstract
Noise Contrastive Estimation (NCE) is a powerful parameter estimation method for log-linear models, which avoids calculation of the partition function or its derivatives at each training step, a computationally demanding step in many cases. It is closely related to negative sampling methods, now widely used in NLP. This paper considers NCE-based estimation of conditional models. Conditional models are frequently encountered in practice; however there has not been a rigorous theoretical analysis of NCE in this setting, and we will argue there are subtle but important questions when generalizing NCE to the conditional case. In particular, we analyze two variants of NCE for conditional models: one based on a classification objective, the other based on a ranking objective. We show that the ranking-based variant of NCE gives consistent parameter estimates under weaker assumptions than the classification-based method; we analyze the statistical efficiency of the ranking-based and classification-based variants of NCE; finally we describe experiments on synthetic data and language modeling showing the effectiveness and trade-offs of both methods.
arXiv commentTo appear in EMNLP2018
References in corpus (4)
Cited by in corpus (7)
- On Mutual Information Maximization for Representation Learning
- Unsupervised State Representation Learning in Atari
- Multi-label Contrastive Predictive Coding
- Simpler, Faster, Stronger: Breaking The log-K Curse On Contrastive Learners With FlatNCE
- Noise Contrastive Meta-Learning for Conditional Density Estimation using Kernel Mean Embeddings
- Predictive Coding for Locally-Linear Control
- Energy-Inspired Models: Learning with Sampler-Induced Distributions