Resolving pronouns to their referents has long been studied as a fundamental natural language understanding problem. Previous works on pronoun coreference resolution (PCR) mostly focus on resolving pronouns to mentions in text while ignoring the exophoric scenario. Exophoric pronouns are common in daily communications, where speakers may directly use pronouns to refer to some objects present in the environment without introducing the objects first. Although such objects are not mentioned in the dialogue text, they can often be disambiguated by the general topics of the dialogue. Motivated by this, we propose to jointly leverage the local context and global topics of dialogues to solve the out-of-text PCR problem. Extensive experiments demonstrate the effectiveness of adding topic regularization for resolving exophoric pronouns.
This report presents the results of a PDE-based binarization model for degraded document images. The model utilizes an edge and binary source term in its formulation. Results indicate effectiveness for document images with bleed-through and faded text and stains to a lesser extent.
Conditional Generative Adversarial Networks (cGANs) extend the standard unconditional GAN framework to learning joint data-label distributions from samples, and have been established as powerful generative models capable of generating high-fidelity imagery. A challenge of training such a model lies in properly infusing class information into its generator and discriminator. For the discriminator, class conditioning can be achieved by either (1) directly incorporating labels as input or (2) involving labels in an auxiliary classification loss. In this paper, we show that the former directly aligns the class-conditioned fake-and-real data distributions $P(\text{image}|\text{class})$ ({\em data matching}), while the latter aligns data-conditioned class distributions $P(\text{class}|\text{image})$ ({\em label matching}). Although class separability does not directly translate to sample quality and becomes a burden if classification itself is intrinsically difficult, the discriminator cannot provide useful guidance for the generator if features of distinct classes are mapped to the same point and thus become inseparable. Motivated by this intuition, we propose a Dual Projection GAN (P2GAN) model that learns to balance between {\em data matching} and {\em label matching}. We then propose an improved cGAN model with Auxiliary Classification that directly aligns the fake and real conditionals $P(\text{class}|\text{image})$ by minimizing their $f$-divergence. Experiments on a synthetic Mixture of Gaussian (MoG) dataset and a variety of real-world datasets including CIFAR100, ImageNet, and VGGFace2 demonstrate the efficacy of our proposed models.
Code-switching, the interleaving of two or more languages within a sentence or discourse is pervasive in multilingual societies. Accurate language models for code-switched text are critical for NLP tasks. State-of-the-art data-intensive neural language models are difficult to train well from scarce language-labeled code-switched text. A potential solution is to use deep generative models to synthesize large volumes of realistic code-switched text. Although generative adversarial networks and variational autoencoders can synthesize plausible monolingual text from continuous latent space, they cannot adequately address code-switched text, owing to their informal style and complex interplay between the constituent languages. We introduce VACS, a novel variational autoencoder architecture specifically tailored to code-switching phenomena. VACS encodes to and decodes from a two-level hierarchical representation, which models syntactic contextual signals in the lower level, and language switching signals in the upper layer. Sampling representations from the prior and decoding them produced well-formed, diverse code-switched sentences. Extensive experiments show that using synthetic code-switched text with natural monolingual data results in significant (33.06%) drop in perplexity.
Voice Activity Detection (VAD) is a fundamental preprocessing step in automatic speech recognition. This is especially true within the broadcast industry where a wide variety of audio materials and recording conditions are encountered. Based on previous studies which indicate that xvector embeddings can be applied to a diverse set of audio classification tasks, we investigate the suitability of x-vectors in discriminating speech from noise. We find that the proposed x-vector based VAD system achieves the best reported score in detecting clean speech on AVA-Speech, whilst retaining robust VAD performance in the presence of noise and music. Furthermore, we integrate the x-vector based VAD system into an existing STT pipeline and compare its performance on multiple broadcast datasets against a baseline system with WebRTC VAD. Crucially, our proposed x-vector based VAD improves the accuracy of STT transcription on real-world broadcast audio
Recently, neural natural language models have attained state-of-the-art performance on a wide variety of tasks, but the high performance can result from superficial, surface-level cues (Bender and Koller, 2020; Niven and Kao, 2020). These surface cues, as the ``shortcuts'' inherent in the datasets, do not contribute to the *task-specific information* (TSI) of the classification tasks. While it is essential to look at the model performance, it is also important to understand the datasets. In this paper, we consider this question: Apart from the information introduced by the shortcut features, how much task-specific information is required to classify a dataset? We formulate this quantity in an information-theoretic framework. While this quantity is hard to compute, we approximate it with a fast and stable method. TSI quantifies the amount of linguistic knowledge modulo a set of predefined shortcuts -- that contributes to classifying a sample from each dataset. This framework allows us to compare across datasets, saying that, apart from a set of ``shortcut features'', classifying each sample in the Multi-NLI task involves around 0.4 nats more TSI than in the Quora Question Pair.
This paper presents a grounded language-image pre-training (GLIP) model for learning object-level, language-aware, and semantic-rich visual representations. GLIP unifies object detection and phrase grounding for pre-training. The unification brings two benefits: 1) it allows GLIP to learn from both detection and grounding data to improve both tasks and bootstrap a good grounding model; 2) GLIP can leverage massive image-text pairs by generating grounding boxes in a self-training fashion, making the learned representation semantic-rich. In our experiments, we pre-train GLIP on 27M grounding data, including 3M human-annotated and 24M web-crawled image-text pairs. The learned representations demonstrate strong zero-shot and few-shot transferability to various object-level recognition tasks. 1) When directly evaluated on COCO and LVIS (without seeing any images in COCO during pre-training), GLIP achieves 49.8 AP and 26.9 AP, respectively, surpassing many supervised baselines. 2) After fine-tuned on COCO, GLIP achieves 60.8 AP on val and 61.5 AP on test-dev, surpassing prior SoTA. 3) When transferred to 13 downstream object detection tasks, a 1-shot GLIP rivals with a fully-supervised Dynamic Head. Code will be released at https://github.com/microsoft/GLIP.
Spoken language understanding (SLU) tasks involve mapping from speech audio signals to semantic labels. Given the complexity of such tasks, good performance might be expected to require large labeled datasets, which are difficult to collect for each new task and domain. However, recent advances in self-supervised speech representations have made it feasible to consider learning SLU models with limited labeled data. In this work we focus on low-resource spoken named entity recognition (NER) and address the question: Beyond self-supervised pre-training, how can we use external speech and/or text data that are not annotated for the task? We draw on a variety of approaches, including self-training, knowledge distillation, and transfer learning, and consider their applicability to both end-to-end models and pipeline (speech recognition followed by text NER model) approaches. We find that several of these approaches improve performance in resource-constrained settings beyond the benefits from pre-trained representations alone. Compared to prior work, we find improved F1 scores of up to 16%. While the best baseline model is a pipeline approach, the best performance when using external data is ultimately achieved by an end-to-end model. We provide detailed comparisons and analyses, showing for example that end-to-end models are able to focus on the more NER-specific words.
Video interpolation is the task that synthesizes the intermediate frame given two consecutive frames. Most of the previous studies have focused on appropriate frame warping operations and refinement modules for the warped frames. These studies have been conducted on natural videos having only continuous motions. However, many practical videos contain a lot of discontinuous motions, such as chat windows, watermarks, GUI elements, or subtitles. We propose three techniques to expand the concept of transition between two consecutive frames to address these issues. First is a new architecture that can separate continuous and discontinuous motion areas. We also propose a novel data augmentation strategy called figure-text mixing (FTM) to make our model learn more general scenarios. Finally, we propose loss functions to give supervisions of the discontinuous motion areas with the data augmentation. We collected a special dataset consisting of some mobile games and chatting videos. We show that our method significantly improves the interpolation qualities of the videos on the special dataset. Moreover, our model outperforms the state-of-the-art methods for natural video datasets containing only continuous motions, such as DAVIS and UCF101.
Automatic spoken instruction understanding (SIU) of the controller-pilot conversations in the air traffic control (ATC) requires not only recognizing the words and semantics of the speech but also determining the role of the speaker. However, few of the published works on the automatic understanding systems in air traffic communication focus on speaker role identification (SRI). In this paper, we formulate the SRI task of controller-pilot communication as a binary classification problem. Furthermore, the text-based, speech-based, and speech and text based multi-modal methods are proposed to achieve a comprehensive comparison of the SRI task. To ablate the impacts of the comparative approaches, various advanced neural network architectures are applied to optimize the implementation of text-based and speech-based methods. Most importantly, a multi-modal speaker role identification network (MMSRINet) is designed to achieve the SRI task by considering both the speech and textual modality features. To aggregate modality features, the modal fusion module is proposed to fuse and squeeze acoustic and textual representations by modal attention mechanism and self-attention pooling layer, respectively. Finally, the comparative approaches are validated on the ATCSpeech corpus collected from a real-world ATC environment. The experimental results demonstrate that all the comparative approaches are worked for the SRI task, and the proposed MMSRINet shows the competitive performance and robustness than the other methods on both seen and unseen data, achieving 98.56%, and 98.08% accuracy, respectively.