The growing security threat of microarchitectural attacks underlines the importance of robust security sensors and detection mechanisms at the hardware level. While there are studies on runtime detection of cache attacks, a generic model to consider the broad range of existing and future attacks is missing. Unfortunately, previous approaches only consider either a single attack variant, e.g. Prime+Probe, or specific victim applications such as cryptographic implementations. Furthermore, the state-of-the art anomaly detection methods are based on coarse-grained statistical models, which are not successful to detect anomalies in a large-scale real world systems. Thanks to the memory capability of advanced Recurrent Neural Networks (RNNs) algorithms, both short and long term dependencies can be learned more accurately. Therefore, we propose FortuneTeller, which for the first time leverages the superiority of RNNs to learn complex execution patterns and detects unseen microarchitectural attacks in real world systems. FortuneTeller models benign workload pattern from a microarchitectural standpoint in an unsupervised fashion, and then, it predicts how upcoming benign executions are supposed to behave. Potential attacks and malicious behaviors will be detected automatically, when there is a discrepancy between the predicted execution pattern and the runtime observation. We implement FortuneTeller based on the available hardware performance counters on Intel processors and it is trained with 10 million samples obtained from benign applications. For the first time, the latest attacks such as Meltdown, Spectre, Rowhammer and Zombieload are detected with one trained model and without observing these attacks during the training. We show that FortuneTeller achieves F-score of 0.9970.
In this paper, we propose a new experimental protocol and use it to benchmark the data efficiency --- performance as a function of training set size --- of two deep learning algorithms, convolutional neural networks (CNNs) and hierarchical information-preserving graph-based slow feature analysis (HiGSFA), for tasks in classification and transfer learning scenarios. The algorithms are trained on different-sized subsets of the MNIST and Omniglot data sets. HiGSFA outperforms standard CNN networks when the models are trained on 50 and 200 samples per class for MNIST classification. In other cases, the CNNs perform better. The results suggest that there are cases where greedy, locally optimal bottom-up learning is equally or more powerful than global gradient-based learning.
While recent advances in deep learning have significantly advanced the state of the art for vessel detection in color fundus (CF) images, the success for detecting vessels in fluorescein angiography (FA) has been stymied due to the lack of labeled ground truth datasets. We propose a novel pipeline to detect retinal vessels in FA images using deep neural networks that reduces the effort required for generating labeled ground truth data by combining two key components: cross-modality transfer and human-in-the-loop learning. The cross-modality transfer exploits concurrently captured CF and fundus FA images. Binary vessels maps are first detected from CF images with a pre-trained neural network and then are geometrically registered with and transferred to FA images via robust parametric chamfer alignment to a preliminary FA vessel detection obtained with an unsupervised technique. Using the transferred vessels as initial ground truth labels for deep learning, the human-in-the-loop approach progressively improves the quality of the ground truth labeling by iterating between deep-learning and labeling. The approach significantly reduces manual labeling effort while increasing engagement. We highlight several important considerations for the proposed methodology and validate the performance on three datasets. Experimental results demonstrate that the proposed pipeline significantly reduces the annotation effort and the resulting deep learning methods outperform prior existing FA vessel detection methods by a significant margin. A new public dataset, RECOVERY-FA19, is introduced that includes high-resolution ultra-widefield images and accurately labeled ground truth binary vessel maps.
Transport protocols use congestion control to avoid overloading a network. Nowadays, different congestion control variants exist that influence performance. Studying their use is thus relevant, but it is hard to identify which variant is used. While passive identification approaches exist, these require detailed domain knowledge and often also rely on outdated assumptions about how congestion control operates and what data is accessible. We present DeePCCI, a passive, deep learning-based congestion control identification approach which does not need any domain knowledge other than training traffic of a congestion control variant. By only using packet arrival data, it is also directly applicable to encrypted (transport header) traffic. DeePCCI is therefore more easily extendable and can also be used with QUIC.
Segmentation has been a major task in neuroimaging. A large number of automated methods have been developed for segmenting healthy and diseased brain tissues. In recent years, deep learning techniques have attracted a lot of attention as a result of their high accuracy in different segmentation problems. We present a new deep learning based segmentation method, DeepMRSeg, that can be applied in a generic way to a variety of segmentation tasks. The proposed architecture combines recent advances in the field of biomedical image segmentation and computer vision. We use a modified UNet architecture that takes advantage of multiple convolution filter sizes to achieve multi-scale feature extraction adaptive to the desired segmentation task. Importantly, our method operates on minimally processed raw MRI scan. We validated our method on a wide range of segmentation tasks, including white matter lesion segmentation, segmentation of deep brain structures and hippocampus segmentation. We provide code and pre-trained models to allow researchers apply our method on their own datasets.
State of the art music recommender systems mainly rely on either matrix factorization-based collaborative filtering approaches or deep learning architectures. Deep learning models usually use metadata for content-based filtering or predict the next user interaction by learning from temporal sequences of user actions. Despite advances in deep learning for song recommendation, none has taken advantage of the sequential nature of songs by learning sequence models that are based on content. Aside from the importance of prediction accuracy, other significant aspects are important, such as explainability and solving the cold start problem. In this work, we propose a hybrid deep learning model, called "SeER", that uses collaborative filtering (CF) and deep learning sequence models on the MIDI content of songs for recommendation in order to provide more accurate personalized recommendations; solve the item cold start problem; and generate a relevant explanation for a song recommendation. Our evaluation experiments show promising results compared to state of the art baseline and hybrid song recommender systems in terms of ranking evaluation. Moreover, based on proposed tests for offline validation, we show that our personalized explanations capture properties that are in accordance with the user's preferences.
Deep learning has recently become very popular on account of its incredible success in many complex data-driven applications, such as image classification and speech recognition. The database community has worked on data-driven applications for many years, and therefore should be playing a lead role in supporting this new wave. However, databases and deep learning are different in terms of both techniques and applications. In this paper, we discuss research problems at the intersection of the two fields. In particular, we discuss possible improvements for deep learning systems from a database perspective, and analyze database applications that may benefit from deep learning techniques.
Generative neural network is a new category of neural networks and it has been widely utilized in applications such as content generation, unsupervised learning, segmentation and pose estimation. It typically involves massive computing-intensive deconvolution operations that cannot be fitted to conventional neural network processors directly. However, prior works mainly investigated specialized hardware architectures through intensive hardware modifications to the existing deep learning processors to accelerate deconvolution together with the convolution. In contrast, this work proposes a novel deconvolution implementation with a software approach and enables fast and efficient deconvolution execution on the legacy deep learning processors. Our proposed method reorganizes the computation of deconvolution and allows the deep learning processors to treat it as the standard convolution by splitting the original deconvolution filters into multiple small filters. Compared to prior acceleration schemes, the implemented acceleration scheme achieves 2.41x - 4.34x performance speedup and reduces the energy consumption by 27.7% - 54.5% on a set of realistic benchmarks. In addition, we also applied the deconvolution computing approach to the off-the-shelf commodity deep learning processors. The performance of deconvolution also exhibits significant performance speedup over prior deconvolution implementations.
Nowadays, users open multiple accounts on social media platforms and e-commerce sites, expressing their personal preferences on different domains. However, users' behaviors change across domains, depending on the content that users interact with, such as movies, music, clothing and retail products. In this paper, we propose an adaptive deep learning strategy for cross-domain recommendation, referred to as ADC. We design a neural architecture and formulate a cross-domain loss function, to compute the non-linearity in user preferences across domains and transfer the knowledge of users' multiple behaviors, accordingly. In addition, we introduce an efficient algorithm for cross-domain loss balancing which directly tunes gradient magnitudes and adapts the learning rates based on the domains' complexities/scales when training the model via backpropagation. In doing so, ADC controls and adjusts the contribution of each domain when optimizing the model parameters. Our experiments on six publicly available cross-domain recommendation tasks demonstrate the effectiveness of the proposed ADC model over other state-of-the-art methods. Furthermore, we study the effect of the proposed adaptive deep learning strategy and show that ADC can well balance the impact of the domains with different complexities.
In this paper we test the use of a deep learning approach to automatically count Wandering Albatrosses in Very High Resolution (VHR) satellite imagery. We use a dataset of manually labelled imagery provided by the British Antarctic Survey to train and develop our methods. We employ a U-Net architecture, designed for image segmentation, to simultaneously classify and localise potential albatrosses. We aid training with the use of the Focal Loss criterion, to deal with extreme class imbalance in the dataset. Initial results achieve peak precision and recall values of approximately 80%. Finally we assess the model's performance in relation to inter-observer variation, by comparing errors against an image labelled by multiple observers. We conclude model accuracy falls within the range of human counters. We hope that the methods will streamline the analysis of VHR satellite images, enabling more frequent monitoring of a species which is of high conservation concern.
In the brain, the structure of a network of neurons defines how these neurons implement the computations that underlie the mind and the behavior of animals and humans. Provided that we can describe the network of neurons as a graph, we can employ methods from graph theory to investigate its structure or use cellular automata to mathematically assess its function. Although, software for the analysis of graphs and cellular automata are widely available. Graph extraction from the image of networks of brain cells remains difficult. Nervous tissue is heterogeneous, and differences in anatomy may reflect relevant differences in function. Here we introduce a deep learning based toolbox to extracts graphs from images of brain tissue. This toolbox provides an easy-to-use framework allowing system neuroscientists to generate graphs based on images of brain tissue by combining methods from image processing, deep learning, and graph theory. The goals are to simplify the training and usage of deep learning methods for computer vision and facilitate its integration into graph extraction pipelines. In this way, the toolbox provides an alternative to the required laborious manual process of tracing, sorting and classifying. We expect to democratize the machine learning methods to a wider community of users beyond the computer vision experts and improve the time-efficiency of graph extraction from large brain image datasets, which may lead to further understanding of the human mind.
Rare diseases affecting 350 million individuals are commonly associated with delay in diagnosis or misdiagnosis. To improve those patients' outcome, rare disease detection is an important task for identifying patients with rare conditions based on longitudinal medical claims. In this paper, we present a deep learning method for detecting patients with exocrine pancreatic insufficiency (EPI) (a rare disease). The contribution includes 1) a large longitudinal study using 7 years medical claims from 1.8 million patients including 29,149 EPI patients, 2) a new deep learning model using generative adversarial networks (GANs) to boost rare disease class, and also leveraging recurrent neural networks to model patient sequence data, 3) an accurate prediction with 0.56 PR-AUC which outperformed benchmark models in terms of precision and recall.
An important step of seismic data processing is removing noise, including interference due to simultaneous and blended sources, from the recorded data. Traditional methods are time-consuming to apply as they often require manual choosing of parameters to obtain good results. We use deep learning, with a U-net model incorporating a ResNet architecture pretrained on ImageNet and further trained on synthetic seismic data, to perform this task. The method is applied to common offset gathers, with adjacent offset gathers of the gather being denoised provided as additional input channels. Here we show that this approach leads to a method that removes noise from several datasets recorded in different parts of the world with moderate success. We find that providing three adjacent offset gathers on either side of the gather being denoised is most effective. As this method does not require parameters to be chosen, it is more automated than traditional methods.
The introduction of deep learning and transfer learning techniques in fields such as computer vision allowed a leap forward in the accuracy of image classification tasks. Currently there is only limited use of such techniques in neuroscience. The challenge of using deep learning methods to successfully train models in neuroscience, lies in the complexity of the information that is processed, the availability of data and the cost of producing sufficient high quality annotations. Inspired by its application in computer vision, we introduce transfer learning on electrophysiological data to enable training a model with limited amounts of data. Our method was tested on the dataset of the BCI competition IV 2a and compared to the top results that were obtained using traditional machine learning techniques. Using our DL model we outperform the top result of the competition by 33%. We also explore transferability of knowledge between trained models over different experiments, called inter-experimental transfer learning. This reduces the amount of required data even further and is especially useful when few subjects are available. This method is able to outperform the standard deep learning methods used in the BCI competition IV 2b approaches by 18%. In this project we propose a method that can produce reliable electroencephalography (EEG) signal classification, based on modest amounts of training data through the use of transfer learning.
We review three recent deep learning based methods for action recognition and present a brief comparative analysis of the methods from a neurophyisiological point of view. We posit that there are some analogy between the three presented deep learning based methods and some of the existing hypotheses regarding the functioning of human brain.
The generative capabilities of deep learning neural networks (DNNs) have been attracting increasing attention for both the remarkable artifacts they produce, but also because of the vast conceptual difference between how they are programmed and what they do. DNNs are 'black boxes' where high-level behavior is not explicitly programmed, but emerges from the complex interactions of thousands or millions of simple computational elements. Their behavior is often described in anthropomorphic terms that can be misleading, seem magical, or stoke fears of an imminent singularity in which machines become 'more' than human. In this paper, we examine 5 distinct behavioral characteristics associated with creativity, and provide an example of a mechanisms from generative deep learning architectures that give rise to each these characteristics. All 5 emerge from machinery built for purposes other than the creative characteristics they exhibit, mostly classification. These mechanisms of creative generative capabilities thus demonstrate a deep kinship to computational perceptual processes. By understanding how these different behaviors arise, we hope to on one hand take the magic out of anthropomorphic descriptions, but on the other, to build a deeper appreciation of machinic forms of creativity on their own terms that will allow us to nurture their further development.
In this paper, we propose a deep learning framework based on randomized neural network. In particular, inspired by the principles of Random Vector Functional Link (RVFL) network, we present a deep RVFL network (dRVFL) with stacked layers. The parameters of the hidden layers of the dRVFL are randomly generated within a suitable range and kept fixed while the output weights are computed using the closed form solution as in a standard RVFL network. We also propose an ensemble deep network (edRVFL) that can be regarded as a marriage of ensemble learning with deep learning. Unlike traditional ensembling approaches that require training several models independently from scratch, edRVFL is obtained by training a single dRVFL network once. Both dRVFL and edRVFL frameworks are generic and can be used with any RVFL variant. To illustrate this, we integrate the deep learning networks with a recently proposed sparse-pretrained RVFL (SP-RVFL). Extensive experiments on benchmark datasets from diverse domains show the superior performance of our proposed deep RVFL networks.
Warmth and competence represent the fundamental traits in social judgment that determine emotional reactions and behavioral intentions towards social targets. This research investigates whether an algorithm can learn visual representations of social categorization and accurately predict human perceivers' impressions of warmth and competence in face images. In addition, this research unravels which areas of a face are important for the classification of warmth and competence. We use Deep Convolutional Neural Networks to extract features from face images and the Gradient-weighted Class Activation Mapping (Grad CAM) method to understand the importance of face regions for the classification. Given a single face image the trained algorithm could correctly predict warmth impressions with an accuracy of about 90% and competence impressions with an accuracy of about 80%. The findings have implications for the automated processing of faces and the design of artificial characters.
While camera and LiDAR processing have been revolutionized since the introduction of deep learning, radar processing still relies on classical tools. In this paper, we introduce a deep learning approach for radar processing, working directly with the radar complex data. To overcome the lack of radar labeled data, we rely in training only on the radar calibration data and introduce new radar augmentation techniques. We evaluate our method on the radar 4D detection task and demonstrate superior performance compared to the classical approaches while keeping real-time performance. Applying deep learning on radar data has several advantages such as eliminating the need for an expensive radar calibration process each time and enabling classification of the detected objects with almost zero-overhead.
In this paper, we propose a novel method for generating a synthetic dataset obeying Gaussian distribution. Compared to the commonly used benchmark datasets with unknown distribution, the synthetic dataset has an explicit distribution, i.e., Gaussian distribution. Meanwhile, it has the same characteristics as the benchmark dataset MNIST. As a result, we can easily apply Deep Neural Networks (DNNs) on the synthetic dataset. This synthetic dataset provides a novel experimental tool to verify the proposed theories of deep learning.