In this work we present Google's submission to the BioASQ 7 biomedical question answering (QA) task (specifically Task 7b, Phase B). The core of our systems are based on BERT QA models, specifically the model of \cite{alberti2019bert}. In this report, and via our submissions, we aimed to investigate two research questions. We start by studying how domain portable are QA systems that have been pre-trained and fine-tuned on general texts, e.g., Wikipedia. We measure this via two submissions. The first is a non-adapted model that uses a public pre-trained BERT model and is fine-tuned on the Natural Questions data set \cite{kwiatkowski2019natural}. The second system takes this non-adapted model and fine-tunes it with the BioASQ training data. Next, we study the impact of error propagation in end-to-end retrieval and QA systems. Again we test this via two submissions. The first uses human annotated relevant documents and snippets as input to the model and the second predicted documents and snippets. Our main findings are that domain specific fine-tuning can benefit Biomedical QA. However, the biggest quality bottleneck is at the retrieval stage, where we see large drops in metrics -- over 10pts absolute -- when using non gold inputs to the QA model.
Existing literature on Question Answering (QA) mostly focuses on algorithmic novelty, data augmentation, or increasingly large pre-trained language models like XLNet and RoBERTa. Additionally, a lot of systems on the QA leaderboards do not have associated research documentation in order to successfully replicate their experiments. In this paper, we outline these algorithmic components such as Attention-over-Attention, coupled with data augmentation and ensembling strategies that have shown to yield state-of-the-art results on benchmark datasets like SQuAD, even achieving super-human performance. Contrary to these prior results, when we evaluate on the recently proposed Natural Questions benchmark dataset, we find that an incredibly simple approach of transfer learning from BERT outperforms the previous state-of-the-art system trained on 4 million more examples than ours by 1.9 F1 points. Adding ensembling strategies further improves that number by 2.3 F1 points.
BERT (Bidirectional Encoder Representations from Transformers) and related pre-trained Transformers have provided large gains across many language understanding tasks, achieving a new state-of-the-art (SOTA). BERT is pre-trained on two auxiliary tasks: Masked Language Model and Next Sentence Prediction. In this paper we introduce a new pre-training task inspired by reading comprehension to better align the pre-training from memorization to understanding. Span Selection Pre-Training (SSPT) poses cloze-like training instances, but rather than draw the answer from the model's parameters, it is selected from a relevant passage. We find significant and consistent improvements over both BERT-BASE and BERT-LARGE on multiple reading comprehension (MRC) datasets. Specifically, our proposed model has strong empirical evidence as it obtains SOTA results on Natural Questions, a new benchmark MRC dataset, outperforming BERT-LARGE by 3 F1 points on short answer prediction. We also show significant impact in HotpotQA, improving answer prediction F1 by 4 points and supporting fact prediction F1 by 1 point and outperforming the previous best system. Moreover, we show that our pre-training approach is particularly effective when training data is limited, improving the learning curve by a large amount.
What does it take to design a machine that learns to answer natural questions about a video? A Video QA system must simultaneously understand language, represent visual content over space-time, and iteratively transform these representations in response to lingual content in the query, and finally arriving at a sensible answer. While recent advances in lingual and visual question answering have enabled sophisticated representations and neural reasoning mechanisms, major challenges in Video QA remain on dynamic grounding of concepts, relations and actions to support the reasoning process. Inspired by the dual-process account of human reasoning, we design a dual process neural architecture, which is composed of a question-guided video processing module (System 1, fast and reactive) followed by a generic reasoning module (System 2, slow and deliberative). System 1 is a hierarchical model that encodes visual patterns about objects, actions and relations in space-time given the textual cues from the question. The encoded representation is a set of high-level visual features, which are then passed to System 2. Here multi-step inference follows to iteratively chain visual elements as instructed by the textual elements. The system is evaluated on the SVQA (synthetic) and TGIF-QA datasets (real), demonstrating competitive results, with a large margin in the case of multi-step reasoning.
Obtaining training data for Question Answering (QA) is time-consuming and resource-intensive, and existing QA datasets are only available for limited domains and languages. In this work, we explore to what extent high quality training data is actually required for Extractive QA, and investigate the possibility of unsupervised Extractive QA. We approach this problem by first learning to generate context, question and answer triples in an unsupervised manner, which we then use to synthesize Extractive QA training data automatically. To generate such triples, we first sample random context paragraphs from a large corpus of documents and then random noun phrases or named entity mentions from these paragraphs as answers. Next we convert answers in context to "fill-in-the-blank" cloze questions and finally translate them into natural questions. We propose and compare various unsupervised ways to perform cloze-to-natural question translation, including training an unsupervised NMT model using non-aligned corpora of natural questions and cloze questions as well as a rule-based approach. We find that modern QA models can learn to answer human questions surprisingly well using only synthetic training data. We demonstrate that, without using the SQuAD training data at all, our approach achieves 56.4 F1 on SQuAD v1 (64.5 F1 when the answer is a Named entity mention), outperforming early supervised models.
Digital traces of our lives are now constantly produced by various connected devices, internet services and interactions. Our actions result in a multitude of heterogeneous data objects, or traces, kept in various locations in the cloud or on local devices. Users have very few tools to organize, understand, and search the digital traces they produce. We propose a simple but flexible data model to aggregate, organize, and find personal information within a collection of a user's personal digital traces. Our model uses as basic dimensions the six questions: what, when, where, who, why, and how. These natural questions model universal aspects of a personal data collection and serve as unifying features of each personal data object, regardless of its source. We propose indexing and search techniques to aid users in searching for their past information in their unified personal digital data sets using our model. Experiments performed over real user data from a variety of data sources such as Facebook, Dropbox, and Gmail show that our approach significantly improves search accuracy when compared with traditional search tools.
This technical note describes a new baseline for the Natural Questions. Our model is based on BERT and reduces the gap between the model F1 scores reported in the original dataset paper and the human upper bound by 30% and 50% relative for the long and short answer tasks respectively. This baseline has been submitted to the official NQ leaderboard at ai.google.com/research/NaturalQuestions. Code, preprocessed data and pretrained model are available at https://github.com/google-research/language/tree/master/language/question_answering/bert_joint.
Generating natural questions from an image is a semantic task that requires using visual and language modality to learn multimodal representations. Images can have multiple visual and language contexts that are relevant for generating questions namely places, captions, and tags. In this paper, we propose the use of exemplars for obtaining the relevant context. We obtain this by using a Multimodal Differential Network to produce natural and engaging questions. The generated questions show a remarkable similarity to the natural questions as validated by a human study. Further, we observe that the proposed approach substantially improves over state-of-the-art benchmarks on the quantitative metrics (BLEU, METEOR, ROUGE, and CIDEr).
Principal components analysis (PCA) is the optimal linear auto-encoder of data, and it is often used to construct features. Enforcing sparsity on the principal components can promote better generalization, while improving the interpretability of the features. We study the problem of constructing optimal sparse linear auto-encoders. Two natural questions in such a setting are: i) Given a level of sparsity, what is the best approximation to PCA that can be achieved? ii) Are there low-order polynomial-time algorithms which can asymptotically achieve this optimal tradeoff between the sparsity and the approximation quality? In this work, we answer both questions by giving efficient low-order polynomial-time algorithms for constructing asymptotically \emph{optimal} linear auto-encoders (in particular, sparse features with near-PCA reconstruction error) and demonstrate the performance of our algorithms on real data.
There has been an explosion of work in the vision & language community during the past few years from image captioning to video transcription, and answering questions about images. These tasks have focused on literal descriptions of the image. To move beyond the literal, we choose to explore how questions about an image are often directed at commonsense inference and the abstract events evoked by objects in the image. In this paper, we introduce the novel task of Visual Question Generation (VQG), where the system is tasked with asking a natural and engaging question when shown an image. We provide three datasets which cover a variety of images from object-centric to event-centric, with considerably more abstract training data than provided to state-of-the-art captioning systems thus far. We train and test several generative and retrieval models to tackle the task of VQG. Evaluation results show that while such models ask reasonable questions for a variety of images, there is still a wide gap with human performance which motivates further work on connecting images with commonsense knowledge and pragmatics. Our proposed task offers a new challenge to the community which we hope furthers interest in exploring deeper connections between vision & language.
Question generation from a knowledge base (KB) is the task of generating questions related to the domain of the input KB. We propose a system for generating fluent and natural questions from a KB, which significantly reduces the human effort by leveraging massive web resources. In more detail, a seed question set is first generated by applying a small number of hand-crafted templates on the input KB, then more questions are retrieved by iteratively forming already obtained questions as search queries into a standard search engine, before finally questions are selected by estimating their fluency and domain relevance. Evaluated by human graders on 500 random-selected triples from Freebase, questions generated by our system are judged to be more fluent than those of \newcite{serban-EtAl:2016:P16-1} by human graders.
Word embedding is a useful approach to capture co-occurrence structures in large text corpora. However, in addition to the text data itself, we often have additional covariates associated with individual corpus documents---e.g. the demographic of the author, time and venue of publication---and we would like the embedding to naturally capture this information. We propose CoVeR, a new tensor decomposition model for vector embeddings with covariates. CoVeR jointly learns a \emph{base} embedding for all the words as well as a weighted diagonal matrix to model how each covariate affects the base embedding. To obtain author or venue-specific embedding, for example, we can then simply multiply the base embedding by the associated transformation matrix. The main advantages of our approach are data efficiency and interpretability of the covariate transformation. Our experiments demonstrate that our joint model learns substantially better covariate-specific embeddings compared to the standard approach of learning a separate embedding for each covariate using only the relevant subset of data, as well as other related methods. Furthermore, CoVeR encourages the embeddings to be "topic-aligned" in that the dimensions have specific independent meanings. This allows our covariate-specific embeddings to be compared by topic, enabling downstream differential analysis. We empirically evaluate the benefits of our algorithm on datasets, and demonstrate how it can be used to address many natural questions about covariate effects. Accompanying code to this paper can be found at http://github.com/kjtian/CoVeR.