Often in medical imaging, it is prohibitively challenging to produce enough boundary annotations to train deep neural networks for accurate tumor segmentation. We propose the use of weak labels about whether an image presents tumor or whether it is absent to extend training over images that lack these annotations. Specifically, we propose a semi-supervised framework that employs unpaired image-to-image translation between two domains, presence vs. absence of cancer, as the unsupervised objective. We conjecture that translation helps segmentation -- both require the target to be separated from the background. We encode images into two codes: one that is common to both domains and one that is unique to the presence domain. Decoding from the common code yields healthy images; decoding with the addition of the unique code produces a residual change to this image that adds cancer. Translation proceeds from presence to absence and vice versa. In the first case, the tumor is re-added to the image and we successfully exploit the residual decoder to also perform segmentation. In the second case, unique codes are sampled, producing a distribution of possible tumors. To validate the method, we created challenging synthetic tasks and tumor segmentation datasets from public BRATS (brain, MRI) and LitS (liver, CT) datasets. We show a clear improvement (0.83 Dice on brain, 0.74 on liver) over baseline semi-supervised training with autoencoding (0.73, 0.66) and a mean teacher approach (0.75, 0.69), demonstrating the ability to generalize from smaller distributions of annotated samples.
This is a report of our lessons learned building acoustic models from 1 Million hours of unlabeled speech, while labeled speech is restricted to 7,000 hours. We employ student/teacher training on unlabeled data, helping scale out target generation in comparison to confidence model based methods, which require a decoder and a confidence model. To optimize storage and to parallelize target generation, we store high valued logits from the teacher model. Introducing the notion of scheduled learning, we interleave learning on unlabeled and labeled data. To scale distributed training across a large number of GPUs, we use BMUF with 64 GPUs, while performing sequence training only on labeled data with gradient threshold compression SGD using 16 GPUs. Our experiments show that extremely large amounts of data are indeed useful; with little hyper-parameter tuning, we obtain relative WER improvements in the 10 to 20% range, with higher gains in noisier conditions.
There is no consensus on the state-of-the-art approach to historical text normalization. Many techniques have been proposed, including rule-based methods, distance metrics, character-based statistical machine translation, and neural encoder--decoder models, but studies have used different datasets, different evaluation methods, and have come to different conclusions. This paper presents the largest study of historical text normalization done so far. We critically survey the existing literature and report experiments on eight languages, comparing systems spanning all categories of proposed normalization techniques, analysing the effect of training data quantity, and using different evaluation methods. The datasets and scripts are made publicly available.
Zero-shot sketch-based image retrieval (ZS-SBIR) is a task of cross-domain image retrieval from a natural image gallery with free-hand sketch under a zero-shot scenario. Previous works mostly focus on a generative approach that takes a highly abstract and sparse sketch as input and then synthesizes the corresponding natural image. However, the intrinsic visual sparsity and large intra-class variance of the sketch make the learning of the conditional decoder more difficult and hence achieve unsatisfactory retrieval performance. In this paper, we propose a novel stacked semantic-guided network to address the unique characteristics of sketches in ZS-SBIR. Specifically, we devise multi-layer feature fusion networks that incorporate different intermediate feature representation information in a deep neural network to alleviate the intrinsic sparsity of sketches. In order to improve visual knowledge transfer from seen to unseen classes, we elaborate a coarse-to-fine conditional decoder that generates coarse-grained category-specific corresponding features first (taking auxiliary semantic information as conditional input) and then generates fine-grained instance-specific corresponding features (taking sketch representation as conditional input). Furthermore, regression loss and classification loss are utilized to preserve the semantic and discriminative information of the synthesized features respectively. Extensive experiments on the large-scale Sketchy dataset and TU-Berlin dataset demonstrate that our proposed approach outperforms state-of-the-art methods by more than 20\% in retrieval performance.
Models trained for classification often assume that all testing classes are known while training. As a result, when presented with an unknown class during testing, such closed-set assumption forces the model to classify it as one of the known classes. However, in a real world scenario, classification models are likely to encounter such examples. Hence, identifying those examples as unknown becomes critical to model performance. A potential solution to overcome this problem lies in a class of learning problems known as open-set recognition. It refers to the problem of identifying the unknown classes during testing, while maintaining performance on the known classes. In this paper, we propose an open-set recognition algorithm using class conditioned auto-encoders with novel training and testing methodology. In contrast to previous methods, training procedure is divided in two sub-tasks, 1. closed-set classification and, 2. open-set identification (i.e. identifying a class as known or unknown). Encoder learns the first task following the closed-set classification training pipeline, whereas decoder learns the second task by reconstructing conditioned on class identity. Furthermore, we model reconstruction errors using the Extreme Value Theory of statistical modeling to find the threshold for identifying known/unknown class samples. Experiments performed on multiple image classification datasets show proposed method performs significantly better than state of the art.
Machine learning (ML) has progressed rapidly during the past decade and the major factor that drives such development is the unprecedented large-scale data. As data generation is a continuous process, this leads to ML model owners updating their models frequently with newly-collected data in an online learning scenario. In consequence, if an ML model is queried with the same set of data samples at two different points in time, it will provide different results. In this paper, we investigate whether the change in the output of a black-box ML model before and after being updated can leak information of the dataset used to perform the update, namely the updating set. This constitutes a new attack surface against black-box ML models and such information leakage may compromise the intellectual property and data privacy of the ML model owner. We propose four attacks following an encoder-decoder formulation, which allows inferring diverse information of the updating set. Our new attacks are facilitated by state-of-the-art deep learning techniques. In particular, we propose a hybrid generative model (CBM-GAN) that is based on generative adversarial networks (GANs) but includes a reconstructive loss that allows reconstructing accurate samples. Our experiments show that the proposed attacks achieve strong performance.
Reading irregular scene text of arbitrary shape in natural images is still a challenging problem, despite the progress made recently. Many existing approaches incorporate sophisticated network structures to handle various shapes, use extra annotations for stronger supervision, or employ hard-to-train recurrent neural networks for sequence modeling. In this work, we propose a simple yet strong approach for scene text recognition. With no need to convert input images to sequence representations, we directly connect two-dimensional CNN features to an attention-based sequence decoder which guided by holistic representation. The holistic representation can guide the attention-based decoder focus on more accurate area. As no recurrent module is adopted, our model can be trained in parallel. It achieves 1.5x to 9.4x acceleration to backward pass and 1.3x to 7.9x acceleration to forward pass, compared with the RNN counterparts. The proposed model is trained with only word-level annotations. With this simple design, our method achieves state-of-the-art or competitive recognition performance on the evaluated regular and irregular scene text benchmark datasets.
We are interested in the approximation of partial differential equations with a data-driven approach based on the reduced basis method and machine learning. We suppose that the phenomenon of interest can be modeled by a parametrized partial differential equation, but that the value of the physical parameters is unknown or difficult to be directly measured. Our method allows to estimate fields of interest, for instance temperature of a sample of material or velocity of a fluid, given data at a handful of points in the domain. We propose to accomplish this task with a neural network embedding a reduced basis solver as exotic activation function in the last layer. The reduced basis solver accounts for the underlying physical phenomenonon and it is constructed from snapshots obtained from randomly selected values of the physical parameters during an expensive offline phase. The same full order solutions are then employed for the training of the neural network. As a matter of fact, the chosen architecture resembles an asymmetric autoencoder in which the decoder is the reduced basis solver and as such it does not contain trainable parameters. The resulting latent space of our autoencoder includes parameter-dependent quantities feeding the reduced basis solver, which -- depending on the considered partial differential equation -- are the values of the physical parameters themselves or the affine decomposition coefficients of the differential operators.
Descriptive comments play a crucial role in the software engineering process. They decrease development time, enable better bug detection, and facilitate the reuse of previously written code. However, comments are commonly the last of a software developer's priorities and are thus either insufficient or missing entirely. Automatic source code summarization may therefore have the ability to significantly improve the software development process. We introduce a novel encoder-decoder model that summarizes source code, effectively writing a comment to describe the code's functionality. We make two primary innovations beyond current source code summarization models. First, our encoder is fully language-agnostic and requires no complex input preprocessing. Second, our decoder has an open vocabulary, enabling it to predict any word, even ones not seen in training. We demonstrate results comparable to state-of-the-art methods on a single-language data set and provide the first results on a data set consisting of multiple programming languages.
In this work, we study abstractive text summarization by exploring different models such as LSTM-encoder-decoder with attention, pointer-generator networks, coverage mechanisms, and transformers. Upon extensive and careful hyperparameter tuning we compare the proposed architectures against each other for the abstractive text summarization task. Finally, as an extension of our work, we apply our text summarization model as a feature extractor for a fake news detection task where the news articles prior to classification will be summarized and the results are compared against the classification using only the original news text. keywords: LSTM, encoder-deconder, abstractive text summarization, pointer-generator, coverage mechanism, transformers, fake news detection
This paper addresses the issue of building a part-based representation of a dataset of images. More precisely, we look for a non-negative, sparse decomposition of the images on a reduced set of atoms, in order to unveil a morphological and interpretable structure of the data. Additionally, we want this decomposition to be computed online for any new sample that is not part of the initial dataset. Therefore, our solution relies on a sparse, non-negative auto-encoder where the encoder is deep (for accuracy) and the decoder shallow (for interpretability). This method compares favorably to the state-of-the-art online methods on two datasets (MNIST and Fashion MNIST), according to classical metrics and to a new one we introduce, based on the invariance of the representation to morphological dilation.
Overcoming the visual barrier and developing "see-through vision" has been one of mankind's long-standing dreams. Unlike visible light, Radio Frequency (RF) signals penetrate opaque obstructions and reflect highly off humans. This paper establishes a deep-learning model that can be trained to reconstruct continuous video of a 15-point human skeleton even through visual occlusion. The training process adopts a student/teacher learning procedure inspired by the Feynman learning technique, in which video frames and RF data are first collected simultaneously using a co-located setup containing an optical camera and an RF antenna array transceiver. Next, the video frames are processed with a computer-vision-based gait analysis "teacher" module to generate ground-truth human skeletons for each frame. Then, the same type of skeleton is predicted from corresponding RF data using a "student" deep-learning model consisting of a Residual Convolutional Neural Network (CNN), Region Proposal Network (RPN), and Recurrent Neural Network with Long-Short Term Memory (LSTM) that 1) extracts spatial features from RF images, 2) detects all people present in a scene, and 3) aggregates information over many time-steps, respectively. The model is shown to both accurately and completely predict the pose of humans behind visual obstruction solely using RF signals. Primary academic contributions include the novel many-to-many imaging methodology, unique integration of RPN and LSTM networks, and original training pipeline.
Learning-based lossy image compression usually involves the joint optimization of rate-distortion performance. Most existing methods adopt spatially invariant bit length allocation and incorporate discrete entropy approximation to constrain compression rate. Nonetheless, the information content is spatially variant, where the regions with complex and salient structures generally are more essential to image compression. Taking the spatial variation of image content into account, this paper presents a content-weighted encoder-decoder model, which involves an importance map subnet to produce the importance mask for locally adaptive bit rate allocation. Consequently, the summation of importance mask can thus be utilized as an alternative of entropy estimation for compression rate control. Furthermore, the quantized representations of the learned code and importance map are still spatially dependent, which can be losslessly compressed using arithmetic coding. To compress the codes effectively and efficiently, we propose a trimmed convolutional network to predict the conditional probability of quantized codes. Experiments show that the proposed method can produce visually much better results, and performs favorably in comparison with deep and traditional lossy image compression approaches.
This paper proposes a novel fault diagnosis approach based on generative adversarial networks (GAN) for imbalanced industrial time series where normal samples are much larger than failure cases. We combine a well-designed feature extractor with GAN to help train the whole network. Aimed at obtaining data distribution and hidden pattern in both original distinguishing features and latent space, the encoder-decoder-encoder three-sub-network is employed in GAN, based on Deep Convolution Generative Adversarial Networks (DCGAN) but without Tanh activation layer and only trained on normal samples. In order to verify the validity and feasibility of our approach, we test it on rolling bearing data from Case Western Reserve University and further verify it on data collected from our laboratory. The results show that our proposed approach can achieve excellent performance in detecting faulty by outputting much larger evaluation scores.
Unsupervised learning is becoming more and more important recently. As one of its key components, the autoencoder (AE) aims to learn a latent feature representation of data which is more robust and discriminative. However, most AE based methods only focus on the reconstruction within the encoder-decoder phase, which ignores the inherent relation of data, i.e., statistical and geometrical dependence, and easily causes overfitting. In order to deal with this issue, we propose an Exclusivity Enhanced (EE) unsupervised feature learning approach to improve the conventional AE. To the best of our knowledge, our research is the first to utilize such exclusivity concept to cooperate with feature extraction within AE. Moreover, in this paper we also make some improvements to the stacked AE structure especially for the connection of different layers from decoders, this could be regarded as a weight initialization trial. The experimental results show that our proposed approach can achieve remarkable performance compared with other related methods.
In the task of machine translation, context information is one of the important factor. But considering the context information model dose not proposed. The paper propose a new model which can integrate context information and make translation. In this paper, we create a new model based Encoder Decoder model. When translating current sentence, the model integrates output from preceding encoder with current encoder. The model can consider context information and the result score is higher than existing model.
We propose a neural network for unsupervised anomaly detection with a novel robust subspace recovery layer (RSR layer). This layer seeks to extract the underlying subspace from a latent representation of the given data and removes outliers that lie away from this subspace. It is used within an autoencoder. The encoder maps the data into a latent space, from which the RSR layer extracts the subspace. The decoder then smoothly maps back the underlying subspace to a "manifold" close to the original inliers. Inliers and outliers are distinguished according to the distances between the original and mapped positions (small for inliers and large for outliers). Extensive numerical experiments with both image and document datasets demonstrate state-of-the-art precision and recall.
Automatic segmentation of brain Magnetic Resonance Imaging (MRI) images is one of the vital steps for quantitative analysis of brain for further inspection. In this paper, NeuroNet has been adopted to segment the brain tissues (white matter (WM), grey matter (GM) and cerebrospinal fluid (CSF)) which uses Residual Network (ResNet) in encoder and Fully Convolution Network (FCN) in the decoder. To achieve the best performance, various hyper-parameters have been tuned, while, network parameters (kernel and bias) were initialized using the NeuroNet pre-trained model. Different pre-processing pipelines have also been introduced to get a robust trained model. The model has been trained and tested on IBSR18 data-set. To validate the research outcome, performance was measured quantitatively using Dice Similarity Coefficient (DSC) and is reported on average as 0.84 for CSF, 0.94 for GM, and 0.94 for WM. The outcome of the research indicates that for the IBSR18 data-set, pre-processing and proper tuning of hyper-parameters for NeuroNet model have improvement in DSC for the brain tissue segmentation.
Variational Autoencoders (VAEs) provide a theoretically-backed and popular framework for deep generative models. However, learning a VAE from data poses still unanswered theoretical questions and considerable practical challenges. In this work, we propose an alternative framework for generative modeling that is simpler, easier to train, and deterministic, yet has many of the advantages of VAEs. We observe that sampling a stochastic encoder in a Gaussian VAE can be interpreted as simply injecting noise into the input of a deterministic decoder. We investigate how substituting this kind of stochasticity, with other explicit and implicit regularization schemes, can lead to an equally smooth and meaningful latent space without forcing it to conform to an arbitrarily chosen prior. To retrieve a generative mechanism to sample new data, we introduce an ex-post density estimation step that can be readily applied also to existing VAEs, improving their sample quality. We show, in a rigorous empirical study, that the proposed regularized deterministic autoencoders are able to generate samples that are comparable to, or better than, those of VAEs and more powerful alternatives when applied to images as well as to structured data such as molecules. \footnote{An implementation is available at: \url{https://github.com/ParthaEth/Regularized_autoencoders-RAE-}}
We investigated the training of a shared model for both text-to-speech (TTS) and voice conversion (VC) tasks. We propose using an extended model architecture of Tacotron, that is a multi-source sequence-to-sequence model with a dual attention mechanism as the shared model for both the TTS and VC tasks. This model can accomplish these two different tasks respectively according to the type of input. An end-to-end speech synthesis task is conducted when the model is given text as the input while a sequence-to-sequence voice conversion task is conducted when it is given the speech of a source speaker as the input. Waveform signals are generated by using WaveNet, which is conditioned by using a predicted mel-spectrogram. We propose jointly training a shared model as a decoder for a target speaker that supports multiple sources. Listening experiments show that our proposed multi-source encoder-decoder model can efficiently achieve both the TTS and VC tasks.