Research papers and code for "Hao Tang":
Segments that span contiguous parts of inputs, such as phonemes in speech, named entities in sentences, actions in videos, occur frequently in sequence prediction problems. Segmental models, a class of models that explicitly hypothesizes segments, have allowed the exploration of rich segment features for sequence prediction. However, segmental models suffer from slow decoding, hampering the use of computationally expensive features. In this thesis, we introduce discriminative segmental cascades, a multi-pass inference framework that allows us to improve accuracy by adding higher-order features and neural segmental features while maintaining efficiency. We also show that instead of including more features to obtain better accuracy, segmental cascades can be used to speed up training and decoding. Segmental models, similarly to conventional speech recognizers, are typically trained in multiple stages. In the first stage, a frame classifier is trained with manual alignments, and then in the second stage, segmental models are trained with manual alignments and the out- puts of the frame classifier. However, obtaining manual alignments are time-consuming and expensive. We explore end-to-end training for segmental models with various loss functions, and show how end-to-end training with marginal log loss can eliminate the need for detailed manual alignments. We draw the connections between the marginal log loss and a popular end-to-end training approach called connectionist temporal classification. We present a unifying framework for various end-to-end graph search-based models, such as hidden Markov models, connectionist temporal classification, and segmental models. Finally, we discuss possible extensions of segmental models to large-vocabulary sequence prediction tasks.

* Thesis
Click to Read Paper and Get Code
Recurrent neural networks have been the dominant models for many speech and language processing tasks. However, we understand little about the behavior and the class of functions recurrent networks can realize. Moreover, the heuristics used during training complicate the analyses. In this paper, we study recurrent networks' ability to learn long-term dependency in the context of speech recognition. We consider two decoding approaches, online and batch decoding, and show the classes of functions to which the decoding approaches correspond. We then draw a connection between batch decoding and a popular training approach for recurrent networks, truncated backpropagation through time. Changing the decoding approach restricts the amount of past history recurrent networks can use for prediction, allowing us to analyze their ability to remember. Empirically, we utilize long-term dependency in subphonetic states, phonemes, and words, and show how the design decisions, such as the decoding approach, lookahead, context frames, and consecutive prediction, characterize the behavior of recurrent networks. Finally, we draw a connection between Markov processes and vanishing gradients. These results have implications for studying the long-term dependency in speech data and how these properties are learned by recurrent networks.

Click to Read Paper and Get Code
Acoustics-to-word models are end-to-end speech recognizers that use words as targets without relying on pronunciation dictionaries or graphemes. These models are notoriously difficult to train due to the lack of linguistic knowledge. It is also unclear how the amount of training data impacts the optimization and generalization of such models. In this work, we study the optimization and generalization of acoustics-to-word models under different amounts of training data. In addition, we study three types of inductive bias, leveraging a pronunciation dictionary, word boundary annotations, and constraints on word durations. We find that constraining word durations leads to the most improvement. Finally, we analyze the word embedding space learned by the model, and find that the space has a structure dominated by the pronunciation of words. This suggests that the contexts of words, instead of their phonetic structure, should be the future focus of inductive bias in acoustics-to-word models.

Click to Read Paper and Get Code
Pulmonary nodule detection, false positive reduction and segmentation represent three of the most common tasks in the computeraided analysis of chest CT images. Methods have been proposed for eachtask with deep learning based methods heavily favored recently. However training deep learning models to solve each task separately may be sub-optimal - resource intensive and without the benefit of feature sharing. Here, we propose a new end-to-end 3D deep convolutional neural net (DCNN), called NoduleNet, to solve nodule detection, false positive reduction and nodule segmentation jointly in a multi-task fashion. To avoid friction between different tasks and encourage feature diversification, we incorporate two major design tricks: 1) decoupled feature maps for nodule detection and false positive reduction, and 2) a segmentation refinement subnet for increasing the precision of nodule segmentation. Extensive experiments on the large-scale LIDC dataset demonstrate that the multi-task training is highly beneficial, improving the nodule detection accuracy by 10.27%, compared to the baseline model trained to only solve the nodule detection task. We also carry out systematic ablation studies to highlight contributions from each of the added components. Code is available at https://github.com/uci-cbcl/NoduleNet.

* Accepted to MICCAI 2019
Click to Read Paper and Get Code
Pulmonary lobe segmentation is an important task for pulmonary disease related Computer Aided Diagnosis systems (CADs). Classical methods for lobe segmentation rely on successful detection of fissures and other anatomical information such as the location of blood vessels and airways. With the success of deep learning in recent years, Deep Convolutional Neural Network (DCNN) has been widely applied to analyze medical images like Computed Tomography (CT) and Magnetic Resonance Imaging (MRI), which, however, requires a large number of ground truth annotations. In this work, we release our manually labeled 50 CT scans which are randomly chosen from the LUNA16 dataset and explore the use of deep learning on this task. We propose pre-processing CT image by cropping region that is covered by the convex hull of the lungs in order to mitigate the influence of noise from outside the lungs. Moreover, we design a hybrid loss function with dice loss to tackle extreme class imbalance issue and focal loss to force model to focus on voxels that are hard to be discriminated. To validate the robustness and performance of our proposed framework trained with a small number of training examples, we further tested our model on CT scans from an independent dataset. Experimental results show the robustness of the proposed approach, which consistently improves performance across different datasets by a maximum of $5.87\%$ as compared to a baseline model.

* 2019 IEEE 16th International Symposium on Biomedical Imaging (ISBI 2019)
Click to Read Paper and Get Code
Gatherings of thousands to millions of people frequently occur for an enormous variety of events, and automated counting of these high-density crowds is useful for safety, management, and measuring significance of an event. In this work, we show that the regularly accepted labeling scheme of crowd density maps for training deep neural networks is less effective than our alternative inverse k-nearest neighbor (i$k$NN) maps, even when used directly in existing state-of-the-art network structures. We also provide a new network architecture MUD-i$k$NN, which uses multi-scale upsampling via transposed convolutions to take full advantage of the provided i$k$NN labeling. This upsampling combined with the i$k$NN maps further improves crowd counting accuracy. Our new network architecture performs favorably in comparison with the state-of-the-art. However, our labeling and upsampling techniques are generally applicable to existing crowd counting architectures.

Click to Read Paper and Get Code
Pulmonary nodule detection using low-dose Computed Tomography (CT) is often the first step in lung disease screening and diagnosis. Recently, algorithms based on deep convolutional neural nets have shown great promise for automated nodule detection. Most of the existing deep learning nodule detection systems are constructed in two steps: a) nodule candidates screening and b) false positive reduction, using two different models trained separately. Although it is commonly adopted, the two-step approach not only imposes significant resource overhead on training two independent deep learning models, but also is sub-optimal because it prevents cross-talk between the two. In this work, we present an end-to-end framework for nodule detection, integrating nodule candidate screening and false positive reduction into one model, trained jointly. We demonstrate that the end-to-end system improves the performance by 3.88\% over the two-step approach, while at the same time reducing model complexity by one third and cutting inference time by 3.6 fold. Code will be made publicly available.

* 2019 IEEE 16th International Symposium on Biomedical Imaging (ISBI 2019)
Click to Read Paper and Get Code
In this work, we generalize semi-supervised generative adversarial networks (GANs) from classification problems to regression problems. In the last few years, the importance of improving the training of neural networks using semi-supervised training has been demonstrated for classification problems. With probabilistic classification being a subset of regression problems, this generalization opens up many new possibilities for the use of semi-supervised GANs as well as presenting an avenue for a deeper understanding of how they function. We first demonstrate the capabilities of semi-supervised regression GANs on a toy dataset which allows for a detailed understanding of how they operate in various circumstances. This toy dataset is used to provide a theoretical basis of the semi-supervised regression GAN. We then apply the semi-supervised regression GANs to the real-world application of age estimation from single images. We perform extensive tests of what accuracies can be achieved with significantly reduced annotated data. Through the combination of the theoretical example and real-world scenario, we demonstrate how semi-supervised GANs can be generalized to regression problems.

Click to Read Paper and Get Code
In this paper, we propose a Convolutional Neural Network (CNN) based speaker recognition model for extracting robust speaker embeddings. The embedding can be extracted efficiently with linear activation in the embedding layer. To understand how the speaker recognition model operates with text-independent input, we modify the structure to extract frame-level speaker embeddings from each hidden layer. We feed utterances from the TIMIT dataset to the trained network and use several proxy tasks to study the networks ability to represent speech input and differentiate voice identity. We found that the networks are better at discriminating broad phonetic classes than individual phonemes. In particular, frame-level embeddings that belong to the same phonetic classes are similar (based on cosine distance) for the same speaker. The frame level representation also allows us to analyze the networks at the frame level, and has the potential for other analyses to improve speaker recognition.

* Accepted at SLT 2018; Supplement materials: https://people.csail.mit.edu/swshon/supplement/slt18.html
Click to Read Paper and Get Code
The state-of-the-art approaches in Generative Adversarial Networks (GANs) are able to learn a mapping function from one image domain to another with unpaired image data. However, these methods often produce artifacts and can only be able to convert low-level information, but fail to transfer high-level semantic part of images. The reason is mainly that generators do not have the ability to detect the most discriminative semantic part of images, which thus makes the generated images with low-quality. To handle the limitation, in this paper we propose a novel Attention-Guided Generative Adversarial Network (AGGAN), which can detect the most discriminative semantic object and minimize changes of unwanted part for semantic manipulation problems without using extra data and models. The attention-guided generators in AGGAN are able to produce attention masks via a built-in attention mechanism, and then fuse the input image with the attention mask to obtain a target image with high-quality. Moreover, we propose a novel attention-guided discriminator which only considers attended regions. The proposed AGGAN is trained by an end-to-end fashion with an adversarial loss, cycle-consistency loss, pixel loss and attention loss. Both qualitative and quantitative results demonstrate that our approach is effective to generate sharper and more accurate images than existing models.

* 8 pages, 7 figures, Accepted to IJCNN 2019
Click to Read Paper and Get Code
Early detection of pulmonary nodules in computed tomography (CT) images is essential for successful outcomes among lung cancer patients. Much attention has been given to deep convolutional neural network (DCNN)-based approaches to this task, but models have relied at least partly on 2D or 2.5D components for inherently 3D data. In this paper, we introduce a novel DCNN approach, consisting of two stages, that is fully three-dimensional end-to-end and utilizes the state-of-the-art in object detection. First, nodule candidates are identified with a U-Net-inspired 3D Faster R-CNN trained using online hard negative mining. Second, false positive reduction is performed by 3D DCNN classifiers trained on difficult examples produced during candidate screening. Finally, we introduce a method to ensemble models from both stages via consensus to give the final predictions. By using this framework, we ranked first of 2887 teams in Season One of Alibaba's 2017 TianChi AI Competition for Healthcare.

* 2018 IEEE 15th International Symposium on Biomedical Imaging (ISBI 2018)
Click to Read Paper and Get Code
The current trend in automatic speech recognition is to leverage large amounts of labeled data to train supervised neural network models. Unfortunately, obtaining data for a wide range of domains to train robust models can be costly. However, it is relatively inexpensive to collect large amounts of unlabeled data from domains that we want the models to generalize to. In this paper, we propose a novel unsupervised adaptation method that learns to synthesize labeled data for the target domain from unlabeled in-domain data and labeled out-of-domain data. We first learn without supervision an interpretable latent representation of speech that encodes linguistic and nuisance factors (e.g., speaker and channel) using different latent variables. To transform a labeled out-of-domain utterance without altering its transcript, we transform the latent nuisance variables while maintaining the linguistic variables. To demonstrate our approach, we focus on a channel mismatch setting, where the domain of interest is distant conversational speech, and labels are only available for close-talking speech. Our proposed method is evaluated on the AMI dataset, outperforming all baselines and bridging the gap between unadapted and in-domain models by over 77% without using any parallel data.

* to appear in Interspeech 2018
Click to Read Paper and Get Code
What is the relationship between sentence representations learned by deep recurrent models against those encoded by the brain? Is there any correspondence between hidden layers of these recurrent models and brain regions when processing sentences? Can these deep models be used to synthesize brain data which can then be utilized in other extrinsic tasks? We investigate these questions using sentences with simple syntax and semantics (e.g., The bone was eaten by the dog.). We consider multiple neural network architectures, including recently proposed ELMo and BERT. We use magnetoencephalography (MEG) brain recording data collected from human subjects when they were reading these simple sentences. Overall, we find that BERT's activations correlate the best with MEG brain data. We also find that the deep network representation can be used to generate brain data from new sentences to augment existing brain data. To the best of our knowledge, this is the first work showing that the MEG brain recording when reading a word in a sentence can be used to distinguish earlier words in the sentence. Our exploration is also the first to use deep neural network representations to generate synthetic brain data and to show that it helps in improving subsequent stimuli decoding task accuracy.

* Association for Computational Linguistics (ACL) 2019
Click to Read Paper and Get Code
Gesture recognition is a hot topic in computer vision and pattern recognition, which plays a vitally important role in natural human-computer interface. Although great progress has been made recently, fast and robust hand gesture recognition remains an open problem, since the existing methods have not well balanced the performance and the efficiency simultaneously. To bridge it, this work combines image entropy and density clustering to exploit the key frames from hand gesture video for further feature extraction, which can improve the efficiency of recognition. Moreover, a feature fusion strategy is also proposed to further improve feature representation, which elevates the performance of recognition. To validate our approach in a "wild" environment, we also introduce two new datasets called HandGesture and Action3D datasets. Experiments consistently demonstrate that our strategy achieves competitive results on Northwestern University, Cambridge, HandGesture and Action3D hand gesture datasets. Our code and datasets will release at https://github.com/Ha0Tang/HandGestureRecognition.

* 11 pages, 3 figures, accepted to NeuroComputing
Click to Read Paper and Get Code
End-to-end training of deep learning-based models allows for implicit learning of intermediate representations based on the final task loss. However, the end-to-end approach ignores the useful domain knowledge encoded in explicit intermediate-level supervision. We hypothesize that using intermediate representations as auxiliary supervision at lower levels of deep networks may be a good way of combining the advantages of end-to-end training and more traditional pipeline approaches. We present experiments on conversational speech recognition where we use lower-level tasks, such as phoneme recognition, in a multitask training approach with an encoder-decoder model for direct character transcription. We compare multiple types of lower-level tasks and analyze the effects of the auxiliary tasks. Our results on the Switchboard corpus show that this approach improves recognition accuracy over a standard encoder-decoder model on the Eval2000 test set.

Click to Read Paper and Get Code
Recent work on discriminative segmental models has shown that they can achieve competitive speech recognition performance, using features based on deep neural frame classifiers. However, segmental models can be more challenging to train than standard frame-based approaches. While some segmental models have been successfully trained end to end, there is a lack of understanding of their training under different settings and with different losses. We investigate a model class based on recent successful approaches, consisting of a linear model that combines segmental features based on an LSTM frame classifier. Similarly to hybrid HMM-neural network models, segmental models of this class can be trained in two stages (frame classifier training followed by linear segmental model weight training), end to end (joint training of both frame classifier and linear weights), or with end-to-end fine-tuning after two-stage training. We study segmental models trained end to end with hinge loss, log loss, latent hinge loss, and marginal log loss. We consider several losses for the case where training alignments are available as well as where they are not. We find that in general, marginal log loss provides the most consistent strong performance without requiring ground-truth alignments. We also find that training with dropout is very important in obtaining good performance with end-to-end training. Finally, the best results are typically obtained by a combination of two-stage training and fine-tuning.

Click to Read Paper and Get Code
Discriminative segmental models, such as segmental conditional random fields (SCRFs) and segmental structured support vector machines (SSVMs), have had success in speech recognition via both lattice rescoring and first-pass decoding. However, such models suffer from slow decoding, hampering the use of computationally expensive features, such as segment neural networks or other high-order features. A typical solution is to use approximate decoding, either by beam pruning in a single pass or by beam pruning to generate a lattice followed by a second pass. In this work, we study discriminative segmental models trained with a hinge loss (i.e., segmental structured SVMs). We show that beam search is not suitable for learning rescoring models in this approach, though it gives good approximate decoding performance when the model is already well-trained. Instead, we consider an approach inspired by structured prediction cascades, which use max-marginal pruning to generate lattices. We obtain a high-accuracy phonetic recognition system with several expensive feature types: a segment neural network, a second-order language model, and second-order phone boundary features.

Click to Read Paper and Get Code
Discriminative segmental models offer a way to incorporate flexible feature functions into speech recognition. However, their appeal has been limited by their computational requirements, due to the large number of possible segments to consider. Multi-pass cascades of segmental models introduce features of increasing complexity in different passes, where in each pass a segmental model rescores lattices produced by a previous (simpler) segmental model. In this paper, we explore several ways of making segmental cascades efficient and practical: reducing the feature set in the first pass, frame subsampling, and various pruning approaches. In experiments on phonetic recognition, we find that with a combination of such techniques, it is possible to maintain competitive performance while greatly reducing decoding, pruning, and training time.

Click to Read Paper and Get Code
We study the problem of recognition of fingerspelled letter sequences in American Sign Language in a signer-independent setting. Fingerspelled sequences are both challenging and important to recognize, as they are used for many content words such as proper nouns and technical terms. Previous work has shown that it is possible to achieve almost 90% accuracies on fingerspelling recognition in a signer-dependent setting. However, the more realistic signer-independent setting presents challenges due to significant variations among signers, coupled with the dearth of available training data. We investigate this problem with approaches inspired by automatic speech recognition. We start with the best-performing approaches from prior work, based on tandem models and segmental conditional random fields (SCRFs), with features based on deep neural network (DNN) classifiers of letters and phonological features. Using DNN adaptation, we find that it is possible to bridge a large part of the gap between signer-dependent and signer-independent performance. Using only about 115 transcribed words for adaptation from the target signer, we obtain letter accuracies of up to 82.7% with frame-level adaptation labels and 69.7% with only word labels.

Click to Read Paper and Get Code