PyTorchKR Papers

최신 논문

총 8,922편
태그·기간 등 상세 필터

카테고리 탐색

로그인 후 이 태그 구독

arXiv:1906.00078

Augmenting C. elegans Microscopic Dataset for Accelerated Pattern Recognition

The detection of cell shape changes in 3D time-lapse images of complex tissues is an important task. However, it is a challenging and tedious task to establish a comprehensive dataset to improve the performance of deep learning models. In the paper, we present a deep learning approach to augment 3D live images of the Caenorhabditis elegans embryo, so that we can further speed up the specific structural pattern recognition. We use an unsupervised training over unlabeled images to generate supplementary datasets for further pattern recognition. Technically, we used Alex-style neural networks in a generative adversarial network framework to generate new datasets that have common features of the C. elegans membrane structure. We also made the dataset available for a broad scientific community.

arXiv:1906.00925공식 코드 · Papers with Code

3D Appearance Super-Resolution with Deep Learning

We tackle the problem of retrieving high-resolution (HR) texture maps of objects that are captured from multiple view points. In the multi-view case, model-based super-resolution (SR) methods have been recently proved to recover high quality texture maps. On the other hand, the advent of deep learning-based methods has already a significant impact on the problem of video and image SR. Yet, a deep learning-based approach to super-resolve the appearance of 3D objects is still missing. The main limitation of exploiting the power of deep learning techniques in the multi-view case is the lack of data. We introduce a 3D appearance SR (3DASR) dataset based on the existing ETH3D [42], SyB3R [31], MiddleBury, and our Collection of 3D scenes from TUM [21], Fountain [51] and Relief [53]. We provide the high- and low-resolution texture maps, the 3D geometric model, images and projection matrices. We exploit the power of 2D learning-based SR methods and design networks suitable for the 3D multi-view case. We incorporate the geometric information by introducing normal maps and further improve the learning process. Experimental results demonstrate that our proposed networks successfully incorporate the 3D geometric information and super-resolve the texture maps.

arXiv:1906.00816공식 코드 · Papers with Code

Bayesian Evidential Deep Learning with PAC Regularization

We propose a novel method for closed-form predictive distribution modeling with neural nets. In quantifying prediction uncertainty, we build on Evidential Deep Learning, which has been impactful as being both simple to implement and giving closed-form access to predictive uncertainty. We employ it to model aleatoric uncertainty and extend it to account also for epistemic uncertainty by converting it to a Bayesian Neural Net. While extending its uncertainty quantification capabilities, we maintain its analytically accessible predictive distribution model by performing progressive moment matching for the first time for approximate weight marginalization. The eventual model introduces a prohibitively large number of hyperparameters for stable training. We overcome this drawback by deriving a vacuous PAC bound that comprises the marginal likelihood of the predictor and a complexity penalty. We observe on regression, classification, and out-of-domain detection benchmarks that our method improves model fit and uncertainty quantification.

arXiv:1906.00617

Perceptual Embedding Consistency for Seamless Reconstruction of Tilewise Style Transfer

Style transfer is a field with growing interest and use cases in deep learning. Recent work has shown Generative Adversarial Networks(GANs) can be used to create realistic images of virtually stained slide images in digital pathology with clinically validated interpretability. Digital pathology images are typically of extremely high resolution, making tilewise analysis necessary for deep learning applications. It has been shown that image generators with instance normalization can cause a tiling artifact when a large image is reconstructed from the tilewise analysis. We introduce a novel perceptual embedding consistency loss significantly reducing the tiling artifact created in the reconstructed whole slide image (WSI). We validate our results by comparing virtually stained slide images with consecutive real stained tissue slide images. We also demonstrate that our model is more robust to contrast, color and brightness perturbations by running comparative sensitivity analysis tests.

arXiv:1906.00537

Incorporating Biological Knowledge with Factor Graph Neural Network for Interpretable Deep Learning

While deep learning has achieved great success in many fields, one common criticism about deep learning is its lack of interpretability. In most cases, the hidden units in a deep neural network do not have a clear semantic meaning or correspond to any physical entities. However, model interpretability and explainability are crucial in many biomedical applications. To address this challenge, we developed the Factor Graph Neural Network model that is interpretable and predictable by combining probabilistic graphical models with deep learning. We directly encode biological knowledge such as Gene Ontology as a factor graph into the model architecture, making the model transparent and interpretable. Furthermore, we devised an attention mechanism that can capture multi-scale hierarchical interactions among biological entities such as genes and Gene Ontology terms. With parameter sharing mechanism, the unrolled Factor Graph Neural Network model can be trained with stochastic depth and generalize well. We applied our model to two cancer genomic datasets to predict target clinical variables and achieved better results than other traditional machine learning and deep learning models. Our model can also be used for gene set enrichment analysis and selecting Gene Ontology terms that are important to target clinical variables.

arXiv:1906.00240

Lung cancer screening with low-dose CT scans using a deep learning approach

Lung cancer is the leading cause of cancer deaths. Early detection through low-dose computed tomography (CT) screening has been shown to significantly reduce mortality but suffers from a high false positive rate that leads to unnecessary diagnostic procedures. Quantitative image analysis coupled to deep learning techniques has the potential to reduce this false positive rate. We conducted a computational analysis of 1449 low-dose CT studies drawn from the National Lung Screening Trial (NLST) cohort. We applied to this cohort our newly developed algorithm, DeepScreener, which is based on a novel deep learning approach. The algorithm, after the training process using about 3000 CT studies, does not require lung nodule annotations to conduct cancer prediction. The algorithm uses consecutive slices and multi-task features to determine whether a nodule is likely to be cancer, and a spatial pyramid to detect nodules at different scales. We find that the algorithm can predict a patient's cancer status from a volumetric lung CT image with high accuracy (78.2%, with area under the Receiver Operating Characteristic curve (AUC) of 0.858). Our preliminary framework ranked 16th of 1972 teams (top 1%) in the Data Science Bowl 2017 (DSB2017) competition, based on the challenge datasets. We report here the application of DeepScreener on an independent NLST test set. This study indicates that the deep learning approach has the potential to significantly reduce the false positive rate in lung cancer screening with low-dose CT scans.

arXiv:1906.00133공식 코드 · Papers with Code

ArcticNet: A Deep Learning Solution to Classify Arctic Wetlands

Arctic environments are rapidly changing under the warming climate. Of particular interest are wetlands, a type of ecosystem that constitutes the most effective terrestrial long-term carbon store. As permafrost thaws, the carbon that was locked in these wetland soils for millennia becomes available for aerobic and anaerobic decomposition, which releases CO2 and CH4, respectively, back to the atmosphere.As CO2 and CH4 are potent greenhouse gases, this transfer of carbon from the land to the atmosphere further contributes to global warming, thereby increasing the rate of permafrost degradation in a positive feedback loop. Therefore, monitoring Arctic wetland health and dynamics is a key scientific task that is also of importance for policy. However, the identification and delineation of these important wetland ecosystems, remain incomplete and often inaccurate. Mapping the extent of Arctic wetlands remains a challenge for the scientific community. Conventional, coarser remote sensing methods are inadequate at distinguishing the diverse and micro-topographically complex non-vascular vegetation that characterize Arctic wetlands, presenting the need for better identification methods. To tackle this challenging problem, we constructed and annotated the first-of-its-kind Arctic Wetland Dataset (AWD). Based on that, we present ArcticNet, a deep neural network that exploits the multi-spectral, high-resolution imagery captured from nanosatellites (Planet Dove CubeSats) with additional DEM from the ArcticDEM project, to semantically label a Arctic study area into six types, in which three Arctic wetland functional types are included. We present multi-fold efforts to handle the arising challenges, including class imbalance, and the choice of fusion strategies. Preliminary results endorse the high promise of ArcticNet, achieving 93.12% in labelling a hold-out set of regions in our Arctic study area.

arXiv:1906.00076

IoT Network Security from the Perspective of Adversarial Deep Learning

Machine learning finds rich applications in Internet of Things (IoT) networks such as information retrieval, traffic management, spectrum sensing, and signal authentication. While there is a surge of interest to understand the security issues of machine learning, their implications have not been understood yet for wireless applications such as those in IoT systems that are susceptible to various attacks due the open and broadcast nature of wireless communications. To support IoT systems with heterogeneous devices of different priorities, we present new techniques built upon adversarial machine learning and apply them to three types of over-the-air (OTA) wireless attacks, namely jamming, spectrum poisoning, and priority violation attacks. By observing the spectrum, the adversary starts with an exploratory attack to infer the channel access algorithm of an IoT transmitter by building a deep neural network classifier that predicts the transmission outcomes. Based on these prediction results, the wireless attack continues to either jam data transmissions or manipulate sensing results over the air (by transmitting during the sensing phase) to fool the transmitter into making wrong transmit decisions in the test phase (corresponding to an evasion attack). When the IoT transmitter collects sensing results as training data to retrain its channel access algorithm, the adversary launches a causative attack to manipulate the input data to the transmitter over the air. We show that these attacks with different levels of energy consumption and stealthiness lead to significant loss in throughput and success ratio in wireless communications for IoT systems. Then we introduce a defense mechanism that systematically increases the uncertainty of the adversary at the inference stage and improves the performance. Results provide new insights on how to attack and defend IoT networks using deep learning.

arXiv:1906.00091공식 코드 · Papers with Code

Deep Learning Recommendation Model for Personalization and Recommendation Systems

With the advent of deep learning, neural network-based recommendation models have emerged as an important tool for tackling personalization and recommendation tasks. These networks differ significantly from other deep learning networks due to their need to handle categorical features and are not well studied or understood. In this paper, we develop a state-of-the-art deep learning recommendation model (DLRM) and provide its implementation in both PyTorch and Caffe2 frameworks. In addition, we design a specialized parallelization scheme utilizing model parallelism on the embedding tables to mitigate memory constraints while exploiting data parallelism to scale-out compute from the fully-connected layers. We compare DLRM against existing recommendation models and characterize its performance on the Big Basin AI platform, demonstrating its usefulness as a benchmark for future algorithmic experimentation and system co-design.

arXiv:1905.13750

Sketch2code: Generating a website from a paper mockup

An early stage of developing user-facing applications is creating a wireframe to layout the interface. Once a wireframe has been created it is given to a developer to implement in code. Developing boiler plate user interface code is time consuming work but still requires an experienced developer. In this dissertation we present two approaches which automates this process, one using classical computer vision techniques, and another using a novel application of deep semantic segmentation networks. We release a dataset of websites which can be used to train and evaluate these approaches. Further, we have designed a novel evaluation framework which allows empirical evaluation by creating synthetic sketches. Our evaluation illustrates that our deep learning approach outperforms our classical computer vision approach and we conclude that deep learning is the most promising direction for future research.

arXiv:1905.13390

Vehicle Detection in Deep Learning

Computer vision is developing rapidly with the support of deep learning techniques. This thesis proposes an advanced vehicle-detection model based on an improvement to classical convolutional neural networks. The advanced model was applied against a vehicle detection benchmark and was built to detect on-road objects. First, we propose a high-level architecture for our advanced model, which utilizes different state-of-the-art deep learning techniques. Then, we utilize the residual neural networks and region proposal network to achieve competitive performance according to the vehicle detection benchmark. Lastly, we describe the developing trend of vehicle detection techniques and the future direction of research.

arXiv:1905.13368

Fast Online "Next Best Offers" using Deep Learning

In this paper, we present iPrescribe, a scalable low-latency architecture for recommending 'next-best-offers' in an online setting. The paper presents the design of iPrescribe and compares its performance for implementations using different real-time streaming technology stacks. iPrescribe uses an ensemble of deep learning and machine learning algorithms for prediction. We describe the scalable real-time streaming technology stack and optimized machine-learning implementations to achieve a 90th percentile recommendation latency of 38 milliseconds. Optimizations include a novel mechanism to deploy recurrent Long Short Term Memory (LSTM) deep learning networks efficiently.

arXiv:1905.13667공식 코드 · Papers with Code

Partial Scanning Transmission Electron Microscopy with Deep Learning

Compressed sensing algorithms are used to decrease electron microscope scan time and electron beam exposure with minimal information loss. Following successful applications of deep learning to compressed sensing, we have developed a two-stage multiscale generative adversarial neural network to complete realistic 512$\times$512 scanning transmission electron micrographs from spiral, jittered gridlike, and other partial scans. For spiral scans and mean squared error based pre-training, this enables electron beam coverage to be decreased by 17.9$\times$ with a 3.8\% test set root mean squared intensity error, and by 87.0$\times$ with a 6.2\% error. Our generator networks are trained on partial scans created from a new dataset of 16227 scanning transmission electron micrographs. High performance is achieved with adaptive learning rate clipping of loss spikes and an auxiliary trainer network. Our source code, new dataset, and pre-trained models have been made publicly available at https://github.com/Jeffrey-Ede/partial-STEM

arXiv:1905.13612

Leveraging Trust and Distrust in Recommender Systems via Deep Learning

The data scarcity of user preferences and the cold-start problem often appear in real-world applications and limit the recommendation accuracy of collaborative filtering strategies. Leveraging the selections of social friends and foes can efficiently face both problems. In this study, we propose a strategy that performs social deep pairwise learning. Firstly, we design a ranking loss function incorporating multiple ranking criteria based on the choice in users, and the choice in their friends and foes to improve the accuracy in the top-k recommendation task. We capture the nonlinear correlations between user preferences and the social information of trust and distrust relationships via a deep learning strategy. In each backpropagation step, we follow a social negative sampling strategy to meet the multiple ranking criteria of our ranking loss function. We conduct comprehensive experiments on a benchmark dataset from Epinions, among the largest publicly available that has been reported in the relevant literature. The experimental results demonstrate that the proposed model beats other state-of-the art methods, attaining an 11.49% average improvement over the most competitive model. We show that our deep learning strategy plays an important role in capturing the nonlinear correlations between user preferences and the social information of trust and distrust relationships, and demonstrate the importance of our social negative sampling strategy on the proposed model.

arXiv:1905.13611

ADMM for Efficient Deep Learning with Global Convergence

Alternating Direction Method of Multipliers (ADMM) has been used successfully in many conventional machine learning applications and is considered to be a useful alternative to Stochastic Gradient Descent (SGD) as a deep learning optimizer. However, as an emerging domain, several challenges remain, including 1) The lack of global convergence guarantees, 2) Slow convergence towards solutions, and 3) Cubic time complexity with regard to feature dimensions. In this paper, we propose a novel optimization framework for deep learning via ADMM (dlADMM) to address these challenges simultaneously. The parameters in each layer are updated backward and then forward so that the parameter information in each layer is exchanged efficiently. The time complexity is reduced from cubic to quadratic in (latent) feature dimensions via a dedicated algorithm design for subproblems that enhances them utilizing iterative quadratic approximations and backtracking. Finally, we provide the first proof of global convergence for an ADMM-based method (dlADMM) in a deep neural network problem under mild conditions. Experiments on benchmark datasets demonstrated that our proposed dlADMM algorithm outperforms most of the comparison methods.

arXiv:1905.13409

Bypassing Backdoor Detection Algorithms in Deep Learning

Deep learning models are vulnerable to various adversarial manipulations of their training data, parameters, and input sample. In particular, an adversary can modify the training data and model parameters to embed backdoors into the model, so the model behaves according to the adversary's objective if the input contains the backdoor features, referred to as the backdoor trigger (e.g., a stamp on an image). The poisoned model's behavior on clean data, however, remains unchanged. Many detection algorithms are designed to detect backdoors on input samples or model parameters, through the statistical difference between the latent representations of adversarial and clean input samples in the poisoned model. In this paper, we design an adversarial backdoor embedding algorithm that can bypass the existing detection algorithms including the state-of-the-art techniques. We design an adaptive adversarial training algorithm that optimizes the original loss function of the model, and also maximizes the indistinguishability of the hidden representations of poisoned data and clean data. This work calls for designing adversary-aware defense mechanisms for backdoor detection.

arXiv:1905.13378

Deep Learning for Distributed Optimization: Applications to Wireless Resource Management

This paper studies a deep learning (DL) framework to solve distributed non-convex constrained optimizations in wireless networks where multiple computing nodes, interconnected via backhaul links, desire to determine an efficient assignment of their states based on local observations. Two different configurations are considered: First, an infinite-capacity backhaul enables nodes to communicate in a lossless way, thereby obtaining the solution by centralized computations. Second, a practical finite-capacity backhaul leads to the deployment of distributed solvers equipped along with quantizers for communication through capacity-limited backhaul. The distributed nature and the nonconvexity of the optimizations render the identification of the solution unwieldy. To handle them, deep neural networks (DNNs) are introduced to approximate an unknown computation for the solution accurately. In consequence, the original problems are transformed to training tasks of the DNNs subject to non-convex constraints where existing DL libraries fail to extend straightforwardly. A constrained training strategy is developed based on the primal-dual method. For distributed implementation, a novel binarization technique at the output layer is developed for quantization at each node. Our proposed distributed DL framework is examined in various network configurations of wireless resource management. Numerical results verify the effectiveness of our proposed approach over existing optimization techniques.

arXiv:1811.0579

LSTM Neural Network for Textual Ngrams

Cognitive neuroscience is the study of how the human brain functions on tasks like decision making, language, perception and reasoning. Deep learning is a class of machine learning algorithms that use neural networks. They are designed to model the responses of neurons in the human brain. Learning can be supervised or unsupervised. Ngram token models are used extensively in language prediction. Ngrams are probabilistic models that are used in predicting the next word or token. They are a statistical model of word sequences or tokens and are called Language Models or Lms. Ngrams are essential in creating language prediction models. We are exploring a broader sandbox ecosystems enabling for AI. Specifically, around Deep learning applications on unstructured content form on the web.

arXiv:1905.13014

Unsupervised Deep Learning for Ultra-reliable and Low-latency Communications

In this paper, we study how to solve resource allocation problems in ultra-reliable and low-latency communications by unsupervised deep learning, which often yield functional optimization problems with quality-of-service (QoS) constraints. We take a joint power and bandwidth allocation problem as an example, which minimizes the total bandwidth required to guarantee the QoS of each user in terms of the delay bound and overall packet loss probability. The global optimal solution is found in a symmetric scenario. A neural network was introduced to find an approximated optimal solution in general scenarios, where the QoS is ensured by using the property that the optimal solution should satisfy as the "supervision signal". Simulation results show that the learning-based solution performs the same as the optimal solution in the symmetric scenario, and can save around 40% bandwidth with respect to the state-of-the-art policy.

arXiv:1905.12827

Deep ensemble learning for Alzheimers disease classification

Ensemble learning use multiple algorithms to obtain better predictive performance than any single one of its constituent algorithms could. With growing popularity of deep learning, researchers have started to ensemble them for various purposes. Few if any, however, has used the deep learning approach as a means to ensemble algorithms. This paper presents a deep ensemble learning framework which aims to harness deep learning algorithms to integrate multisource data and tap the wisdom of experts. At the voting layer, a sparse autoencoder is trained for feature learning to reduce the correlation of attributes and diversify the base classifiers ultimately. At the stacking layer, a nonlinear feature-weighted method based on deep belief networks is proposed to rank the base classifiers which may violate the conditional independence. Neural network is used as meta classifier. At the optimizing layer, under-sampling and threshold-moving are used to cope with cost-sensitive problem. Optimized predictions are obtained based on ensemble of probabilistic predictions by similarity calculation. The proposed deep ensemble learning framework is used for Alzheimers disease classification. Experiments with the clinical dataset from national Alzheimers coordinating center demonstrate that the classification accuracy of our proposed framework is 4% better than 6 well-known ensemble approaches as well as the standard stacking algorithm. Adequate coverage of more accurate diagnostic services can be provided by utilizing the wisdom of averaged physicians. This paper points out a new way to boost the primary care of Alzheimers disease from the view of machine learning.