PyTorchKR Papers

최신 논문

총 8,922편
태그·기간 등 상세 필터

카테고리 탐색

로그인 후 이 태그 구독

arXiv:1909.07930공식 코드 · Papers with Code

Ludwig: a type-based declarative deep learning toolbox

In this work we present Ludwig, a flexible, extensible and easy to use toolbox which allows users to train deep learning models and use them for obtaining predictions without writing code. Ludwig implements a novel approach to deep learning model building based on two main abstractions: data types and declarative configuration files. The data type abstraction allows for easier code and sub-model reuse, and the standardized interfaces imposed by this abstraction allow for encapsulation and make the code easy to extend. Declarative model definition configuration files enable inexperienced users to obtain effective models and increase the productivity of expert users. Alongside these two innovations, Ludwig introduces a general modularized deep learning architecture called Encoder-Combiner-Decoder that can be instantiated to perform a vast amount of machine learning tasks. These innovations make it possible for engineers, scientists from other fields and, in general, a much broader audience to adopt deep learning models for their tasks, concretely helping in its democratization.

arXiv:1909.07817

DeepDriveMD: Deep-Learning Driven Adaptive Molecular Simulations for Protein Folding

Simulations of biological macromolecules play an important role in understanding the physical basis of a number of complex processes such as protein folding. Even with increasing computational power and evolution of specialized architectures, the ability to simulate protein folding at atomistic scales still remains challenging. This stems from the dual aspects of high dimensionality of protein conformational landscapes, and the inability of atomistic molecular dynamics (MD) simulations to sufficiently sample these landscapes to observe folding events. Machine learning/deep learning (ML/DL) techniques, when combined with atomistic MD simulations offer the opportunity to potentially overcome these limitations by: (1) effectively reducing the dimensionality of MD simulations to automatically build latent representations that correspond to biophysically relevant reaction coordinates (RCs), and (2) driving MD simulations to automatically sample potentially novel conformational states based on these RCs. We examine how coupling DL approaches with MD simulations can fold small proteins effectively on supercomputers. In particular, we study the computational costs and effectiveness of scaling DL-coupled MD workflows by folding two prototypical systems, viz., Fs-peptide and the fast-folding variant of the villin head piece protein. We demonstrate that a DL driven MD workflow is able to effectively learn latent representations and drive adaptive simulations. Compared to traditional MD-based approaches, our approach achieves an effective performance gain in sampling the folded states by at least 2.3x. Our study provides a quantitative basis to understand how DL driven MD simulations, can lead to effective performance gains and reduced times to solution on supercomputing resources.

arXiv:1909.07786

High Resolution Forecasting of Heat Waves impacts on Leaf Area Index by Multiscale Multitemporal Deep Learning

Climate change impacts could cause progressive decrease of crop quality and yield, up to harvest failures. In particular, heat waves and other climate extremes can lead to localized food shortages and even threaten food security of communities worldwide. In this study, we apply a deep learning architecture for high resolution forecasting (300 m, 10 days) of the Leaf Area Index (LAI), whose dynamics has been widely used to model the growth phase of crops and impact of heat waves. LAI models can be computed at 0.1 degree spatial resolution with an auto regressive component adjusted with weather conditions, validated with remote sensing measurements. However model actionability is poor in regions of varying terrain morphology at this scale (about 8 km at the Alps latitude). Our deep learning model aims instead at forecasting LAI by training multiscale multitemporal (MSMT) data from the Copernicus Global Land Service (CGLS) project for all Europe at 300m resolution and medium-resolution historical weather data. Further, the deep learning model inputs integrate high-resolution land surface features, known to improve forecasts of agricultural productivity. The historical weather data are then replaced with forecast values to predict LAI values at 10 day horizon on Europe. We propose the MSMT model to develop a high resolution crop-specific warning system for mitigating damage due to heat waves and other extreme events.

arXiv:1909.07782공식 코드 · Papers with Code

Interpolation-Prediction Networks for Irregularly Sampled Time Series

In this paper, we present a new deep learning architecture for addressing the problem of supervised learning with sparse and irregularly sampled multivariate time series. The architecture is based on the use of a semi-parametric interpolation network followed by the application of a prediction network. The interpolation network allows for information to be shared across multiple dimensions of a multivariate time series during the interpolation stage, while any standard deep learning model can be used for the prediction network. This work is motivated by the analysis of physiological time series data in electronic health records, which are sparse, irregularly sampled, and multivariate. We investigate the performance of this architecture on both classification and regression tasks, showing that our approach outperforms a range of baseline and recently proposed models.

arXiv:1909.07729

K-TanH: Efficient TanH For Deep Learning

We propose K-TanH, a novel, highly accurate, hardware efficient approximation of popular activation function TanH for Deep Learning. K-TanH consists of parameterized low-precision integer operations, such as, shift and add/subtract (no floating point operation needed) where parameters are stored in very small look-up tables that can fit in CPU registers. K-TanH can work on various numerical formats, such as, Float32 and BFloat16. High quality approximations to other activation functions, e.g., Sigmoid, Swish and GELU, can be derived from K-TanH. Our AVX512 implementation of K-TanH demonstrates $>5\times$ speed up over Intel SVML, and it is consistently superior in efficiency over other approximations that use floating point arithmetic. Finally, we achieve state-of-the-art Bleu score and convergence results for training language translation model GNMT on WMT16 data sets with approximate TanH obtained via K-TanH on BFloat16 inputs.

arXiv:1909.07632공식 코드 · Papers with Code

Towards Efficient and Secure Delivery of Data for Deep Learning with Privacy-Preserving

Privacy recently emerges as a severe concern in deep learning, that is, sensitive data must be prohibited from being shared with the third party during deep neural network development. In this paper, we propose Morphed Learning (MoLe), an efficient and secure scheme to deliver deep learning data. MoLe has two main components: data morphing and Augmented Convolutional (Aug-Conv) layer. Data morphing allows data providers to send morphed data without privacy information, while Aug-Conv layer helps deep learning developers to apply their networks on the morphed data without performance penalty. MoLe provides stronger security while introducing lower overhead compared to GAZELLE (USENIX Security 2018), which is another method with no performance penalty on the neural network. When using MoLe for VGG-16 network on CIFAR dataset, the computational overhead is only 9% and the data transmission overhead is 5.12%. As a comparison, GAZELLE has computational overhead of 10,000 times and data transmission overhead of 421,000 times. In this setting, the attack success rate of adversary is 7.9 x 10^{-90} for MoLe and 2.9 x 10^{-30} for GAZELLE, respectively.

arXiv:1909.07587

A Hybrid Deep Learning Approach for Diagnosis of the Erythemato-Squamous Disease

The diagnosis of the Erythemato-squamous disease (ESD) is accepted as a difficult problem in dermatology. ESD is a form of skin disease. It generally causes redness of the skin and also may cause loss of skin. They are generally due to genetic or environmental factors. ESD comprises six classes of skin conditions namely, pityriasis rubra pilaris, lichen planus, chronic dermatitis, psoriasis, seboreic dermatitis and pityriasis rosea. The automated diagnosis of ESD can help doctors and dermatologists in reducing the efforts from their end and in taking faster decisions for treatment. The literature is replete with works that used conventional machine learning methods for the diagnosis of ESD. However, there isn't much instances of application of Deep learning for the diagnosis of ESD. In this paper, we propose a novel hybrid deep learning approach i.e. Derm2Vec for the diagnosis of the ESD. Derm2Vec is a hybrid deep learning model that consists of both Autoencoders and Deep Neural Networks. We also apply a conventional Deep Neural Network (DNN) for the classification of ESD. We apply both Derm2Vec and DNN along with other traditional machine learning methods on a real world dermatology dataset. The Derm2Vec method is found to be the best performer (when taking the prediction accuracy into account) followed by DNN and Extreme Gradient Boosting.The mean CV score of Derm2Vec, DNN and Extreme Gradient Boosting are 96.92 percent, 96.65 percent and 95.80 percent respectively.

arXiv:1909.06526

FfDL : A Flexible Multi-tenant Deep Learning Platform

Deep learning (DL) is becoming increasingly popular in several application domains and has made several new application features involving computer vision, speech recognition and synthesis, self-driving automobiles, drug design, etc. feasible and accurate. As a result, large scale on-premise and cloud-hosted deep learning platforms have become essential infrastructure in many organizations. These systems accept, schedule, manage and execute DL training jobs at scale. This paper describes the design, implementation and our experiences with FfDL, a DL platform used at IBM. We describe how our design balances dependability with scalability, elasticity, flexibility and efficiency. We examine FfDL qualitatively through a retrospective look at the lessons learned from building, operating, and supporting FfDL; and quantitatively through a detailed empirical evaluation of FfDL, including the overheads introduced by the platform for various deep learning models, the load and performance observed in a real case study using FfDL within our organization, the frequency of various faults observed including unanticipated faults, and experiments demonstrating the benefits of various scheduling policies. FfDL has been open-sourced.

arXiv:1909.06397

Horizontal Flows and Manifold Stochastics in Geometric Deep Learning

We introduce two constructions in geometric deep learning for 1) transporting orientation-dependent convolutional filters over a manifold in a continuous way and thereby defining a convolution operator that naturally incorporates the rotational effect of holonomy; and 2) allowing efficient evaluation of manifold convolution layers by sampling manifold valued random variables that center around a weighted diffusion mean. Both methods are inspired by stochastics on manifolds and geometric statistics, and provide examples of how stochastic methods -- here horizontal frame bundle flows and non-linear bridge sampling schemes, can be used in geometric deep learning. We outline the theoretical foundation of the two methods, discuss their relation to Euclidean deep networks and existing methodology in geometric deep learning, and establish important properties of the proposed constructions.

arXiv:1909.07046

Identifying Pediatric Vascular Anomalies With Deep Learning

Vascular anomalies, more colloquially known as birthmarks, affect up to 1 in 10 infants. Though many of these lesions self-resolve, some types can result in medical complications or disfigurement without proper diagnosis or management. Accurately diagnosing vascular anomalies is challenging for pediatricians and primary care physicians due to subtle visual differences and similarity to other pediatric dermatologic conditions. This can result in delayed or incorrect referrals for treatment. To address this problem, we developed a convolutional neural network (CNN) to automatically classify images of vascular anomalies and other pediatric skin conditions to aid physicians with diagnosis. We constructed a dataset of 21,681 clinical images, including data collected between 2002-2018 at Seattle Children's hospital as well as five dermatologist-curated online repositories, and built a taxonomy over vascular anomalies and other common pediatric skin lesions. The CNN achieved an average AUC of 0.9731 when ten-fold cross-validation was performed across a taxonomy of 12 classes. The classifier's average AUC and weighted F1 score was 0.9889 and 0.9732 respectively when evaluated on a previously unseen test set of six of these classes. Further, when used as an aid by pediatricians (n = 7), the classifier increased their average visual diagnostic accuracy from 73.10% to 91.67%. The classifier runs in real-time on a smartphone and has the potential to improve diagnosis of these conditions, particularly in resource-limited areas.

arXiv:1909.06929

Unaligned Sequence Similarity Search Using Deep Learning

Gene annotation has traditionally required direct comparison of DNA sequences between an unknown gene and a database of known ones using string comparison methods. However, these methods do not provide useful information when a gene does not have a close match in the database. In addition, each comparison can be costly when the database is large since it requires alignments and a series of string comparisons. In this work we propose a novel approach: using recurrent neural networks to embed DNA or amino-acid sequences in a low-dimensional space in which distances correlate with functional similarity. This embedding space overcomes both shortcomings of the method of aligning sequences and comparing homology. First, it allows us to obtain information about genes which do not have exact matches by measuring their similarity to other ones in the database. If our database is labeled this can provide labels for a query gene as is done in traditional methods. However, even if the database is unlabeled it allows us to find clusters and infer some characteristics of the gene population. In addition, each comparison is much faster than traditional methods since the distance metric is reduced to the Euclidean distance, and thus efficient approximate nearest neighbor algorithms can be used to find the best match. We present results showing the advantage of our algorithm. More specifically we show how our embedding can be useful for both classification tasks when our labels are known, and clustering tasks where our sequences belong to classes which have not been seen before.

arXiv:1909.06842

Benchmarking the Performance and Energy Efficiency of AI Accelerators for AI Training

Deep learning has become widely used in complex AI applications. Yet, training a deep neural network (DNNs) model requires a considerable amount of calculations, long running time, and much energy. Nowadays, many-core AI accelerators (e.g., GPUs and TPUs) are designed to improve the performance of AI training. However, processors from different vendors perform dissimilarly in terms of performance and energy consumption. To investigate the differences among several popular off-the-shelf processors (i.e., Intel CPU, NVIDIA GPU, AMD GPU, and Google TPU) in training DNNs, we carry out a comprehensive empirical study on the performance and energy efficiency of these processors by benchmarking a representative set of deep learning workloads, including computation-intensive operations, classical convolutional neural networks (CNNs), recurrent neural networks (LSTM), Deep Speech 2, and Transformer. Different from the existing end-to-end benchmarks which only present the training time, We try to investigate the impact of hardware, vendor's software library, and deep learning framework on the performance and energy consumption of AI training. Our evaluation methods and results not only provide an informative guide for end-users to select proper AI accelerators, but also expose some opportunities for the hardware vendors to improve their software library.

arXiv:1909.06723공식 코드 · Papers with Code

Natural Language Adversarial Defense through Synonym Encoding

In the area of natural language processing, deep learning models are recently known to be vulnerable to various types of adversarial perturbations, but relatively few works are done on the defense side. Especially, there exists few effective defense method against the successful synonym substitution based attacks that preserve the syntactic structure and semantic information of the original text while fooling the deep learning models. We contribute in this direction and propose a novel adversarial defense method called Synonym Encoding Method (SEM). Specifically, SEM inserts an encoder before the input layer of the target model to map each cluster of synonyms to a unique encoding and trains the model to eliminate possible adversarial perturbations without modifying the network architecture or adding extra data. Extensive experiments demonstrate that SEM can effectively defend the current synonym substitution based attacks and block the transferability of adversarial examples. SEM is also easy and efficient to scale to large models and big datasets.

arXiv:1909.07330

Knowledge Discovery In Nanophotonics Using Geometric Deep Learning

We present here a new approach for using the intelligence aspects of artificial intelligence for knowledge discovery rather than device optimization in electromagnetic (EM) nanostructures. This approach uses training data obtained through full-wave EM simulations of a series of nanostructures to train geometric deep learning algorithms to assess the range of feasible responses as well as the feasibility of a desired response from a class of EM nanostructures. To facilitate the knowledge discovery and reduce the computation complexity, our approach combines the dimensionality reduction technique (using an autoencoder) with convex-hull and one-class support-vector-machine (SVM) algorithms to find the range of the feasible responses in the latent (or the reduced) response space of the EM nanostructure. We show that by using a small set of training instances (compared to all possible structures), our approach can provide better than 95% accuracy in assessing the feasibility of a given response. More importantly, the one-class SVM algorithm can be trained to provide the degree of feasibility (or unfeasibility) of a response from a given nanostructure. This important information can be used to modify the initial structure to an alternative one that can enable an initially unfeasible response. To show the applicability of our approach, we apply it to two important classes of binary metasurfaces (MSs), formed by array of plasmonic nanostructures, and periodic MSs formed by an array of dielectric nanopillars. In addition to theoretical results, we show the experimental results obtained by fabricating several MSs of the second class. Our theoretical and experimental results confirm the unique features of this approach for knowledge discovery in EM nanostructures.

arXiv:1909.06326

Automatic Hip Fracture Identification and Functional Subclassification with Deep Learning

Purpose: Hip fractures are a common cause of morbidity and mortality. Automatic identification and classification of hip fractures using deep learning may improve outcomes by reducing diagnostic errors and decreasing time to operation. Methods: Hip and pelvic radiographs from 1118 studies were reviewed and 3034 hips were labeled via bounding boxes and classified as normal, displaced femoral neck fracture, nondisplaced femoral neck fracture, intertrochanteric fracture, previous ORIF, or previous arthroplasty. A deep learning-based object detection model was trained to automate the placement of the bounding boxes. A Densely Connected Convolutional Neural Network (DenseNet) was trained on a subset of the bounding box images, and its performance evaluated on a held out test set and by comparison on a 100-image subset to two groups of human observers: fellowship-trained radiologists and orthopaedists, and senior residents in emergency medicine, radiology, and orthopaedics. Results: The binary accuracy for fracture of our model was 93.8% (95% CI, 91.3-95.8%), with sensitivity of 92.7% (95% CI, 88.7-95.6%), and specificity 95.0% (95% CI, 91.5-97.3%). Multiclass classification accuracy was 90.4% (95% CI, 87.4-92.9%). When compared to human observers, our model achieved at least expert-level classification under all conditions. Additionally, when the model was used as an aid, human performance improved, with aided resident performance approximating unaided fellowship-trained expert performance. Conclusions: Our deep learning model identified and classified hip fractures with at least expert-level accuracy, and when used as an aid improved human performance, with aided resident performance approximating that of unaided fellowship-trained attendings.

논문 ID: 2018.28333

Pulmonary Artery–Vein Classification in CT Images Using Deep Learning

Recent studies show that pulmonary vascular diseases may specifically affect arteries or veins through different physiologic mechanisms. To detect changes in the two vascular trees, physicians manually analyze the chest computed tomography (CT) image of the patients in search of abnormalities. This process is time consuming, difficult to standardize, and thus not feasible for large clinical studies or useful in real-world clinical decision making. Therefore, automatic separation of arteries and veins in CT images is becoming of great interest, as it may help physicians to accurately diagnose pathological conditions. In this paper, we present a novel, fully automatic approach to classify vessels from chest CT images into arteries and veins. The algorithm follows three main steps: first, a scale-space particles segmentation to isolate vessels; then a 3-D convolutional neural network (CNN) to obtain a first classification of vessels; finally, graph-cuts' optimization to refine the results. To justify the usage of the proposed CNN architecture, we compared different 2-D and 3-D CNNs that may use local information from bronchus- and vessel-enhanced images provided to the network with different strategies. We also compared the proposed CNN approach with a random forests (RFs) classifier. The methodology was trained and evaluated on the superior and inferior lobes of the right lung of 18 clinical cases with noncontrast chest CT scans, in comparison with manual classification. The proposed algorithm achieves an overall accuracy of 94%, which is higher than the accuracy obtained using other CNN architectures and RF. Our method was also validated with contrast-enhanced CT scans of patients with chronic thromboembolic pulmonary hypertension to demonstrate that our model generalizes well to contrast-enhanced modalities. The proposed method outperforms state-of-the-art methods, paving the way for future use of 3-D CNN for artery/vein classification in CT images.

arXiv:1909.06312공식 코드 · Papers with Code

Neural Oblivious Decision Ensembles for Deep Learning on Tabular Data

Nowadays, deep neural networks (DNNs) have become the main instrument for machine learning tasks within a wide range of domains, including vision, NLP, and speech. Meanwhile, in an important case of heterogenous tabular data, the advantage of DNNs over shallow counterparts remains questionable. In particular, there is no sufficient evidence that deep learning machinery allows constructing methods that outperform gradient boosting decision trees (GBDT), which are often the top choice for tabular problems. In this paper, we introduce Neural Oblivious Decision Ensembles (NODE), a new deep learning architecture, designed to work with any tabular data. In a nutshell, the proposed NODE architecture generalizes ensembles of oblivious decision trees, but benefits from both end-to-end gradient-based optimization and the power of multi-layer hierarchical representation learning. With an extensive experimental comparison to the leading GBDT packages on a large number of tabular datasets, we demonstrate the advantage of the proposed NODE architecture, which outperforms the competitors on most of the tasks. We open-source the PyTorch implementation of NODE and believe that it will become a universal framework for machine learning on tabular data.

arXiv:1909.06264

A superpixel-driven deep learning approach for the analysis of dermatological wounds

Background. The image-based identification of distinct tissues within dermatological wounds enhances patients' care since it requires no intrusive evaluations. This manuscript presents an approach, we named QTDU, that combines deep learning models with superpixel-driven segmentation methods for assessing the quality of tissues from dermatological ulcers. Method. QTDU consists of a three-stage pipeline for the obtaining of ulcer segmentation, tissues' labeling, and wounded area quantification. We set up our approach by using a real and annotated set of dermatological ulcers for training several deep learning models to the identification of ulcered superpixels. Results. Empirical evaluations on 179,572 superpixels divided into four classes showed QTDU accurately spot wounded tissues (AUC = 0.986, sensitivity = 0.97, and specificity = 0.974) and outperformed machine-learning approaches in up to 8.2% regarding F1-Score through fine-tuning of a ResNet-based model. Last, but not least, experimental evaluations also showed QTDU correctly quantified wounded tissue areas within a 0.089 Mean Absolute Error ratio. Conclusions. Results indicate QTDU effectiveness for both tissue segmentation and wounded area quantification tasks. When compared to existing machine-learning approaches, the combination of superpixels and deep learning models outperformed the competitors within strong significant levels.

arXiv:1909.06040공식 코드 · Papers with Code

DL2: A Deep Learning-driven Scheduler for Deep Learning Clusters

More and more companies have deployed machine learning (ML) clusters, where deep learning (DL) models are trained for providing various AI-driven services. Efficient resource scheduling is essential for maximal utilization of expensive DL clusters. Existing cluster schedulers either are agnostic to ML workload characteristics, or use scheduling heuristics based on operators' understanding of particular ML framework and workload, which are less efficient or not general enough. In this paper, we show that DL techniques can be adopted to design a generic and efficient scheduler. DL2 is a DL-driven scheduler for DL clusters, targeting global training job expedition by dynamically resizing resources allocated to jobs. DL2 advocates a joint supervised learning and reinforcement learning approach: a neural network is warmed up via offline supervised learning based on job traces produced by the existing cluster scheduler; then the neural network is plugged into the live DL cluster, fine-tuned by reinforcement learning carried out throughout the training progress of the DL jobs, and used for deciding job resource allocation in an online fashion. By applying past decisions made by the existing cluster scheduler in the preparatory supervised learning phase, our approach enables a smooth transition from existing scheduler, and renders a high-quality scheduler in minimizing average training completion time. We implement DL2 on Kubernetes and enable dynamic resource scaling in DL jobs on MXNet. Extensive evaluation shows that DL2 outperforms fairness scheduler (i.e., DRF) by 44.1% and expert heuristic scheduler (i.e., Optimus) by 17.5% in terms of average job completion time.

arXiv:1909.06020

Spectrum Sensing Based on Deep Learning Classification for Cognitive Radios

Spectrum sensing is a key technology for cognitive radios. We present spectrum sensing as a classification problem and propose a sensing method based on deep learning classification. We normalize the received signal power to overcome the effects of noise power uncertainty. We train the model with as many types of signals as possible as well as noise data to enable the trained network model to adapt to untrained new signals. We also use transfer learning strategies to improve the performance for real-world signals. Extensive experiments are conducted to evaluate the performance of this method. The simulation results show that the proposed method performs better than two traditional spectrum sensing methods, i.e., maximum-minimum eigenvalue ratio-based method and frequency domain entropy-based method. In addition, the experimental results of the new untrained signal types show that our method can adapt to the detection of these new signals. Furthermore, the real-world signal detection experiment results show that the detection performance can be further improved by transfer learning. Finally, experiments under colored noise show that our proposed method has superior detection performance under colored noise, while the traditional methods have a significant performance degradation, which further validate the superiority of our method.