Training on only perfect Standard English corpora predisposes pre-trained neural networks to discriminate against minorities from non-standard linguistic backgrounds (e.g., African American Vernacular English, Colloquial Singapore English, etc.). We perturb the inflectional morphology of words to craft plausible and semantically similar adversarial examples that expose these biases in popular NLP models, e.g., BERT and Transformer, and show that adversarially fine-tuning them for a single epoch significantly improves robustness without sacrificing performance on clean data.

The choice of hyper-parameters affects the performance of neural models. While much previous research (Sutskever et al., 2013; Duchi et al., 2011; Kingma and Ba, 2015) focuses on accelerating convergence and reducing the effects of the learning rate, comparatively few papers concentrate on the effect of batch size. In this paper, we analyze how increasing batch size affects gradient direction, and propose to evaluate the stability of gradients with their angle change. Based on our observations, the angle change of gradient direction first tends to stabilize (i.e. gradually decrease) while accumulating mini-batches, and then starts to fluctuate. We propose to automatically and dynamically determine batch sizes by accumulating gradients of mini-batches and performing an optimization step at just the time when the direction of gradients starts to fluctuate. To improve the efficiency of our approach for large models, we propose a sampling approach to select gradients of parameters sensitive to the batch size. Our approach dynamically determines proper and efficient batch sizes during training. In our experiments on the WMT 14 English to German and English to French tasks, our approach improves the Transformer with a fixed 25k batch size by +0.73 and +0.82 BLEU respectively.

In work implicitly includes monolingual and natural language processing, the main promise of cross-lingual signals that constitute a departure multilingual learning is to bridge the digital language from the pure setting. We review existing training divide, to enable access to information and signals as well as other signals that may be technology for the world's 6,900 languages (Ruder of interest for future study (§4). We then discuss et al., 2019). For the purpose of this paper, we methodological issues in UCL (e.g., validation, hyperparameter define "multilingual learning" as learning a common tuning) and propose best evaluation model for two or more languages from raw practices (§5). Finally, we provide a unified outlook text, without any downstream task labels. Common of established research areas (cross-lingual use cases include translation as well as pretraining word embeddings, deep multilingual models and multilingual representations. We will use the term unsupervised machine translation) in UCL (§6), interchangeably with "cross-lingual learning".

Neural networks are known to suffer from catastrophic forgetting when trained on sequential datasets. While there have been numerous attempts to solve this problem for large-scale supervised classification, little has been done to overcome catastrophic forgetting for few-shot classification problems. We demonstrate that the popular gradient-based few-shot meta-learning algorithm Model-Agnostic Meta-Learning (MAML) indeed suffers from catastrophic forgetting and introduce a Bayesian online meta-learning framework that tackles this problem. Our framework incorporates MAML into a Bayesian online learning algorithm with Laplace approximation. This framework enables few-shot classification on a range of sequentially arriving datasets with a single meta-learned model. The experimental evaluations demonstrate that our framework can effectively prevent forgetting in various few-shot classification settings compared to applying MAML sequentially.

In data-parallel synchronous training of deep neural networks, different devices (replicas) run the same program with different partitions of the training batch, but weight update computation is repeated on all replicas, because the weights do not have a batch dimension to partition. This can be a bottleneck for performance and scalability in typical language models with large weights, and models with small per-replica batch size which is typical in large-scale training. This paper presents an approach to automatically shard the weight update computation across replicas with efficient communication primitives and data formatting, using static analysis and transformations on the training computation graph. We show this technique achieves substantial speedups on typical image and language models on Cloud TPUs, requiring no change to model code. This technique helps close the gap between traditionally expensive (ADAM) and cheap (SGD) optimizers, as they will only take a small part of training step time and have similar peak memory usage. It helped us to achieve state-of-the-art training performance in Google's MLPerf 0.6 submission.

Artificial Intelligence (AI) is already re-configuring the world in conspicuous ways. Data drives our global digital ecosystem, and AI technologies reveal patterns in data. Smartphones, smart homes, and smart cities influence how we live and interact, and AI systems are increasingly involved in recruitment decisions, medical diagnoses, and judicial verdicts. Whether this scenario is utopian or dystopian depends on your perspective. The potential risks of AI are enumerated repeatedly.

Variational autoencoders (VAEs) are important tools in end-to-end representation learning. VAEs can capture complex data distributions and have been applied extensively in many natural-language-processing (NLP) tasks. However, a common pitfall in sequence-to-sequence learning with VAEs is the posterior-collapse issue in latent space, wherein the model tends to ignore latent variables when a strong auto-regressive decoder is implemented. In this paper, we propose a principled approach to eliminate this issue by applying a discretized bottleneck in the latent space. Specifically, we impose a shared discrete latent space where each input is learned to choose a combination of shared latent atoms as its latent representation. Compared with VAEs employing continuous latent variables, our model endows more promising capability in modeling underlying semantics of discrete sequences and can thus provide more interpretative latent structures. Empirically, we demonstrate the efficiency and effectiveness of our model on a broad range of tasks, including language modeling, unaligned text style transfer, dialog response generation, and neural machine translation.

Perhaps we can expect an iPhone-like symphonic progression in models here? Many companies are throwing their hat into the translation technology ring. Web translation software is being surpassed by portable, state-of-the-art technology in the form of earpieces, hand-held devices and apps, all of which are enabling users to quickly navigate our multilingual world on-the-go. Most recently, American Airlines announced it is testing interpreter mode for Google Assistant to help communication between their employees and travellers who speak a different language. In recent years, artificial intelligence (AI) has drastically enhanced the accuracy and quality of foreign language translations – allowing machines to help break down language barriers for customer service teams and tourists alike.

The quality of automatic metrics for machine translation has been increasingly called into question, especially for high-quality systems. This paper demonstrates that, while choice of metric is important, the nature of the references is also critical. We study different methods to collect references and compare their value in automated evaluation by reporting correlation with human evaluation for a variety of systems and metrics. Motivated by the finding that typical references exhibit poor diversity, concentrating around translationese language, we develop a paraphrasing task for linguists to perform on existing reference translations, which counteracts this bias. Our method yields higher correlation with human judgment not only for the submissions of WMT 2019 English to German, but also for Back-translation and APE augmented MT output, which have been shown to have low correlation with automatic metrics using standard references. We demonstrate that our methodology improves correlation with all modern evaluation metrics we look at, including embedding-based methods. To complete this picture, we reveal that multi-reference BLEU does not improve the correlation for high quality output, and present an alternative multi-reference formulation that is more effective.

Neural Machine Translation (NMT) methodologies have burgeoned from using simple feed-forward architectures to the state of the art; viz. BERT model. The use cases of NMT models have been broadened from just language translations to conversational agents (chatbots), abstractive text summarization, image captioning, etc. which have proved to be a gem in their respective applications. This paper aims to study the major trends in Neural Machine Translation, the state of the art models in the domain and a high level comparison between them.