We study the use of BERT for non-factoid question-answering, focusing on the passage re-ranking task under varying passage lengths. To this end, we explore the fine-tuning of BERT in different learning-to-rank setups, comprising both point-wise and pair-wise methods, resulting in substantial improvements over the state-of-the-art. We then analyze the effectiveness of BERT for different passage lengths and suggest how to cope with large passages.
Recent advances in deep learning greatly boost the performance of object detection. State-of-the-art methods such as Faster-RCNN, FPN and R-FCN have achieved high accuracy in challenging benchmark datasets. However, these methods require fully annotated object bounding boxes for training, which are incredibly hard to scale up due to the high annotation cost. Weakly-supervised methods, on the other hand, only require image-level labels for training, but the performance is far below their fully-supervised counterparts. In this paper, we propose a semi-supervised large scale fine-grained detection method, which only needs bounding box annotations of a smaller number of coarse-grained classes and image-level labels of large scale fine-grained classes, and can detect all classes at nearly fully-supervised accuracy. We achieve this by utilizing the correlations between coarse-grained and fine-grained classes with shared backbone, soft-attention based proposal re-ranking, and a dual-level memory module. Experiment results show that our methods can achieve close accuracy on object detection to state-of-the-art fully-supervised methods on two large scale datasets, ImageNet and OpenImages, with only a small fraction of fully annotated classes.
A major challenge in matching images and text is that they have intrinsically different data distributions and feature representations. Most existing approaches are based either on embedding or classification, the first one mapping image and text instances into a common embedding space for distance measuring, and the second one regarding image-text matching as a binary classification problem. Neither of these approaches can, however, balance the matching accuracy and model complexity well. We propose a novel framework that achieves remarkable matching performance with acceptable model complexity. Specifically, in the training stage, we propose a novel Multi-modal Tensor Fusion Network (MTFN) to explicitly learn an accurate image-text similarity function with rank-based tensor fusion rather than seeking a common embedding space for each image-text instance. Then, during testing, we deploy a generic Cross-modal Re-ranking (RR) scheme for refinement without requiring additional training procedure. Extensive experiments on two datasets demonstrate that our MTFN-RR consistently achieves the state-of-the-art matching performance with much less time complexity. The implementation code is available at https://github.com/Wangt-CN/MTFN-RR-PyTorch-Code.
We introduce the task of retrieving relevant video moments from a large corpus of untrimmed, unsegmented videos given a natural language query. Our task poses unique challenges as a system must efficiently identify both the relevant videos and localize the relevant moments in the videos. To address these challenges, we propose SpatioTemporal Alignment with Language (STAL), a model that represents a video moment as a set of regions within a series of short video clips and aligns a natural language query to the moment's regions. Our alignment cost compares variable-length language and video features using symmetric squared Chamfer distance, which allows for efficient indexing and retrieval of the video moments. Moreover, aligning language features to regions within a video moment allows for finer alignment compared to methods that extract only an aggregate feature from the entire video moment. We evaluate our approach on two recently proposed datasets for temporal localization of moments in video with natural language (DiDeMo and Charades-STA) extended to our video corpus moment retrieval setting. We show that our STAL re-ranking model outperforms the recently proposed Moment Context Network on all criteria across all datasets on our proposed task, obtaining relative gains of 37% - 118% for average recall and up to 30% for median rank. Moreover, our approach achieves more than 130x faster retrieval and 8x smaller index size with a 1M video corpus in an approximate setting.
This technical report presents an overview of our solution used in the submission to ActivityNet Challenge 2019 Task 1 (\textbf{temporal action proposal generation}) and Task 2 (\textbf{temporal action localization/detection}). Temporal action proposal indicates the temporal intervals containing the actions and plays an important role in temporal action localization. Top-down and bottom-up methods are the two main categories used for proposal generation in the existing literature. In this paper, we devise a novel Multi-Granularity Fusion Network (MGFN) to combine the proposals generated from different frameworks for complementary filtering and confidence re-ranking. Specifically, we consider the diversity comprehensively from multiple perspectives, e.g. the characteristic aspect, the data aspect, the model aspect and the result aspect. Our MGFN achieves the state-of-the-art performance on the temporal action proposal task with 69.85 AUC score and the temporal action localization task with 38.90 mAP on the challenge testing set.
Open book question answering is a type of natural language based QA (NLQA) where questions are expected to be answered with respect to a given set of open book facts, and common knowledge about a topic. Recently a challenge involving such QA, OpenBookQA, has been proposed. Unlike most other NLQA tasks that focus on linguistic understanding, OpenBookQA requires deeper reasoning involving linguistic understanding as well as reasoning with common knowledge. In this paper we address QA with respect to the OpenBookQA dataset and combine state of the art language models with abductive information retrieval (IR), information gain based re-ranking, passage selection and weighted scoring to achieve 72.0% accuracy, an 11.6% improvement over the current state of the art.
Visible-infrared cross-modality person re-identification (VI-ReId) is an essential task for video surveillance in poorly illuminated or dark environments. Despite many recent studies on person re-identification in the visible domain (ReId), there are few studies dealing specifically with VI-ReId. Besides challenges that are common for both ReId and VI-ReId such as pose/illumination variations, background clutter and occlusion, VI-ReId has additional challenges as color information is not available in infrared images. As a result, the performance of VI-ReId systems is typically lower than that of ReId systems. In this work, we propose a four-stream framework to improve VI-ReId performance. We train a separate deep convolutional neural network in each stream using different representations of input images. We expect that different and complementary features can be learned from each stream. In our framework, grayscale and infrared input images are used to train the ResNet in the first stream. In the second stream, RGB and three-channel infrared images (created by repeating the infrared channel) are used. In the remaining two streams, we use local pattern maps as input images. These maps are generated utilizing local Zernike moments transformation. Local pattern maps are obtained from grayscale and infrared images in the third stream and from RGB and three-channel infrared images in the last stream. We improve the performance of the proposed framework by employing a re-ranking algorithm for post-processing. Our results indicate that the proposed framework outperforms current state-of-the-art with a large margin by improving Rank-1/mAP by 29.79%/30.91% on SYSU-MM01 dataset, and by 9.73%/16.36% on RegDB dataset.
Establishing a docker-based replicability infrastructure offers the community a great opportunity: measuring the run time of information retrieval systems. The time required to present query results to a user is paramount to the users satisfaction. Recent advances in neural IR re-ranking models put the issue of query latency at the forefront. They bring a complex trade-off between performance and effectiveness based on a myriad of factors: the choice of encoding model, network architecture, hardware acceleration and many others. The best performing models (currently using the BERT transformer model) run orders of magnitude more slowly than simpler architectures. We aim to broaden the focus of the neural IR community to include performance considerations -- to sustain the practical applicability of our innovations. In this position paper we supply our argument with a case study exploring the performance of different neural re-ranking models. Finally, we propose to extend the OSIRRC docker-based replicability infrastructure with two performance focused benchmark scenarios.
Classical information retrieval (IR) methods, such as query likelihood and BM25, score documents independently w.r.t. each query term, and then accumulate the scores. Assuming query term independence allows precomputing term-document scores using these models---which can be combined with specialized data structures, such as inverted index, for efficient retrieval. Deep neural IR models, in contrast, compare the whole query to the document and are, therefore, typically employed only for late stage re-ranking. We incorporate query term independence assumption into three state-of-the-art neural IR models: BERT, Duet, and CKNRM---and evaluate their performance on a passage ranking task. Surprisingly, we observe no significant loss in result quality for Duet and CKNRM---and a small degradation in the case of BERT. However, by operating on each query term independently, these otherwise computationally intensive models become amenable to offline precomputation---dramatically reducing the cost of query evaluations employing state-of-the-art neural ranking models. This strategy makes it practical to use deep models for retrieval from large collections---and not restrict their usage to late stage re-ranking.
Parallel deep learning architectures like fine-tuned BERT and MT-DNN, have quickly become the state of the art, bypassing previous deep and shallow learning methods by a large margin. More recently, pre-trained models from large related datasets have been able to perform well on many downstream tasks by just fine-tuning on domain-specific datasets . However, using powerful models on non-trivial tasks, such as ranking and large document classification, still remains a challenge due to input size limitations of parallel architecture and extremely small datasets (insufficient for fine-tuning). In this work, we introduce an end-to-end system, trained in a multi-task setting, to filter and re-rank answers in the medical domain. We use task-specific pre-trained models as deep feature extractors. Our model achieves the highest Spearman's Rho and Mean Reciprocal Rank of 0.338 and 0.9622 respectively, on the ACL-BioNLP workshop MediQA Question Answering shared-task.
A common approach for knowledge-base entity search is to consider an entity as a document with multiple fields. Models that focus on matching query terms in different fields are popular choices for searching such entity representations. An instance of such a model is FSDM (Fielded Sequential Dependence Model). We propose to integrate field-level semantic features into FSDM. We use FSDM to retrieve a pool of documents, and then to use semantic field-level features to re-rank those documents. We propose to represent queries as bags of terms as well as bags of entities, and eventually, use their dense vector representation to compute semantic features based on query document similarity. Our proposed re-ranking approach achieves significant improvement in entity retrieval on the DBpedia-Entity (v2) dataset over existing FSDM model. Specifically, for all queries we achieve 2.5% and 1.2% significant improvement in NDCG@10 and NDCG@100, respectively.
In this paper, we describe our solution to the Google Landmark Recognition 2019 Challenge held on Kaggle. Due to the large number of classes, noisy data, imbalanced class sizes, and the presence of a significant amount of distractors in the test set, our method is based mainly on retrieval techniques with both global and local CNN approaches. Our full pipeline, after ensembling the models and applying several steps of re-ranking strategies, scores 0.37606 GAP on the private leaderboard which won the 1st place in the competition.
We analyze different re-ranking algorithms for diversification and show that majority of them are based on maximizing submodular/modular functions from the class of parameterized concave/linear over modular functions. We study the optimality of such algorithms in terms of the `total curvature'. We also show that by adjusting the hyperparameter of the concave/linear composition to trade-off relevance and diversity, if any, one is in fact tuning the `total curvature' of the function for relevance-diversity trade-off.
As one of the quintessence of Chinese traditional culture, couplet compromises two syntactically symmetric clauses equal in length, namely, an antecedent and subsequent clause. Moreover, corresponding characters and phrases at the same position of the two clauses are paired with each other under certain constraints of semantic and/or syntactic relatedness. Automatic couplet generation is recognized as a challenging problem even in the Artificial Intelligence field. In this paper, we comprehensively study on automatic generation of acrostic couplet with the first characters defined by users. The complete couplet generation is mainly divided into three stages, that is, antecedent clause generation pipeline, subsequent clause generation pipeline and clause re-ranker. To realize semantic and/or syntactic relatedness between two clauses, attention-based Sequence-to-Sequence (S2S) neural network is employed. Moreover, to provide diverse couplet candidates for re-ranking, a cluster-based beam search approach is incorporated into the S2S network. Both BLEU metrics and human judgments have demonstrated the effectiveness of our proposed method. Eventually, a mini-program based on this generation system is developed and deployed on Wechat for real users.
We propose a novel method for selecting coherent and diverse responses for a given dialogue context. The proposed method re-ranks response candidates generated from conversational models by using event causality relations between events in a dialogue history and response candidates (e.g., ``be stressed out'' precedes ``relieve stress''). We use distributed event representation based on the Role Factored Tensor Model for a robust matching of event causality relations due to limited event causality knowledge of the system. Experimental results showed that the proposed method improved coherency and dialogue continuity of system responses.
The Google-Landmarks-v2 dataset is the biggest worldwide landmarks dataset characterized by a large magnitude of noisiness and diversity. We present a novel landmark retrieval/recognition system, robust to a noisy and diverse dataset, by our team, smlyaka. Our approach is based on deep convolutional neural networks with metric learning, trained by cosine-softmax based losses. Deep metric learning methods are usually sensitive to noise, and it could hinder to learn a reliable metric. To address this issue, we develop an automated data cleaning system. Besides, we devise a discriminative re-ranking method to address the diversity of the dataset for landmark retrieval. Using our methods, we achieved 1st place in the Google Landmark Retrieval 2019 challenge and 3rd place in the Google Landmark Recognition 2019 challenge on Kaggle.
The conventional paradigm in neural question answering (QA) for narrative content is limited to a two-stage process: first, relevant text passages are retrieved and, subsequently, a neural network for machine comprehension extracts the likeliest answer. However, both stages are largely isolated in the status quo and, hence, information from the two phases is never properly fused. In contrast, this work proposes RankQA: RankQA extends the conventional two-stage process in neural QA with a third stage that performs an additional answer re-ranking. The re-ranking leverages different features that are directly extracted from the QA pipeline, i.e., a combination of retrieval and comprehension features. While our intentionally simple design allows for an efficient, data-sparse estimation, it nevertheless outperforms more complex QA systems by a significant margin: in fact, RankQA achieves state-of-the-art performance on 3 out of 4 benchmark datasets. Furthermore, its performance is especially superior in settings where the size of the corpus is dynamic. Here the answer re-ranking provides an effective remedy against the underlying noise-information trade-off due to a variable corpus size. As a consequence, RankQA represents a novel, powerful, and thus challenging baseline for future research in content-based QA.
Many recommender systems suffer from popularity bias: popular items are recommended frequently while less popular, niche products, are recommended rarely or not at all. However, recommending the ignored products in the `long tail' is critical for businesses as they are less likely to be discovered. In this paper, we introduce a personalized diversification re-ranking approach to increase the representation of less popular items in recommendations while maintaining acceptable recommendation accuracy. Our approach is a post-processing step that can be applied to the output of any recommender system. We show that our approach is capable of managing popularity bias more effectively, compared with an existing method based on regularization. We also examine both new and existing metrics to measure the coverage of long-tail items in the recommendation.
We present a framework for quantifying and mitigating algorithmic bias in mechanisms designed for ranking individuals, typically used as part of web-scale search and recommendation systems. We first propose complementary measures to quantify bias with respect to protected attributes such as gender and age. We then present algorithms for computing fairness-aware re-ranking of results. For a given search or recommendation task, our algorithms seek to achieve a desired distribution of top ranked results with respect to one or more protected attributes. We show that such a framework can be tailored to achieve fairness criteria such as equality of opportunity and demographic parity depending on the choice of the desired distribution. We evaluate the proposed algorithms via extensive simulations over different parameter choices, and study the effect of fairness-aware ranking on both bias and utility measures. We finally present the online A/B testing results from applying our framework towards representative ranking in LinkedIn Talent Search, and discuss the lessons learned in practice. Our approach resulted in tremendous improvement in the fairness metrics (nearly three fold increase in the number of search queries with representative results) without affecting the business metrics, which paved the way for deployment to 100% of LinkedIn Recruiter users worldwide. Ours is the first large-scale deployed framework for ensuring fairness in the hiring domain, with the potential positive impact for more than 630M LinkedIn members.
The bidirectional encoder representations from transformers (BERT) model has recently advanced the state-of-the-art in passage re-ranking. In this paper, we analyze the results produced by a fine-tuned BERT model to better understand the reasons behind such substantial improvements. To this aim, we focus on the MS MARCO passage re-ranking dataset and provide potential reasons for the successes and failures of BERT for retrieval. In more detail, we empirically study a set of hypotheses and provide additional analysis to explain the successful performance of BERT.