Machine Learning as a Service (MLaaS) is enabling a wide range of smart applications on end devices. However, such convenience comes with a cost of privacy because users have to upload their private data to the cloud. This research aims to provide effective and efficient MLaaS such that the cloud server learns nothing about user data and the users cannot infer the proprietary model parameters owned by the server. This work makes the following contributions. First, it unveils the fundamental performance bottleneck of existing schemes due to the heavy permutations in computing linear transformation and the use of communication intensive Garbled Circuits for nonlinear transformation. Second, it introduces an ultra-fast secure MLaaS framework, CHEETAH, which features a carefully crafted secret sharing scheme that runs significantly faster than existing schemes without accuracy loss. Third, CHEETAH is evaluated on the benchmark of well-known, practical deep networks such as AlexNet and VGG-16 on the MNIST and ImageNet datasets. The results demonstrate more than 100x speedup over the fastest GAZELLE (Usenix Security'18), 2000x speedup over MiniONN (ACM CCS'17) and five orders of magnitude speedup over CryptoNets (ICML'16). This significant speedup enables a wide range of practical applications based on privacy-preserved deep neural networks.

We introduce Negative Sampling in Semi-Supervised Learning (NS3L), a simple, fast, easy to tune algorithm for semi-supervised learning (SSL). NS3L is motivated by the success of negative sampling/contrastive estimation. We demonstrate that adding the NS3L loss to state-of-the-art SSL algorithms, such as the Virtual Adversarial Training (VAT), significantly improves upon vanilla VAT and its variant, VAT with Entropy Minimization. By adding the NS3L loss to MixMatch, the current state-of-the-art approach on semi-supervised tasks, we observe significant improvements over vanilla MixMatch. We conduct extensive experiments on the CIFAR10, CIFAR100, SVHN and STL10 benchmark datasets.

Monitoring physiological responses to hemodynamic stress can help in determining appropriate treatment and ensuring good patient outcomes. Physicians' intuition suggests that the human body has a number of physiological response patterns to hemorrhage which escalate as blood loss continues, however the exact etiology and phenotypes of such responses are not well known or understood only at a coarse level. Although previous research has shown that machine learning models can perform well in hemorrhage detection and survival prediction, it is unclear whether machine learning could help to identify and characterize the underlying physiological responses in raw vital sign data. We approach this problem by first transforming the high-dimensional vital sign time series into a tractable, lower-dimensional latent space using a dilated, causal convolutional encoder model trained purely unsupervised. Second, we identify informative clusters in the embeddings. By analyzing the clusters of latent embeddings and visualizing them over time, we hypothesize that the clusters correspond to the physiological response patterns that match physicians' intuition. Furthermore, we attempt to evaluate the latent embeddings using a variety of methods, such as predicting the cluster labels using explainable features.

Prospective display advertising poses a great challenge for large advertising platforms as the strongest predictive signals of users are not eligible to be used in the conversion prediction systems. To that end efforts are made to collect as much information as possible about each user from various data sources and to design powerful models that can capture weaker signals ultimately obtaining good quality of conversion prediction probability estimates. In this study we propose a novel time-aware approach to model heterogeneous sequences of users' activities and capture implicit signals of users' conversion intents. On two real-world datasets we show that our approach outperforms other, previously proposed approaches, while providing interpretability of signal impact to conversion probability.

Reinforcement learning is effective in optimizing policies for recommender systems. Current solutions mostly focus on model-free approaches, which require frequent interactions with a real environment, and thus are expensive in model learning. Offline evaluation methods, such as importance sampling, can alleviate such limitations, but usually request a large amount of logged data and do not work well when the action space is large. In this work, we propose a model-based reinforcement learning solution which models the user-agent interaction for offline policy learning via a generative adversarial network. To reduce bias in the learnt policy, we use the discriminator to evaluate the quality of generated sequences and rescale the generated rewards. Our theoretical analysis and empirical evaluations demonstrate the effectiveness of our solution in identifying patterns from given offline data and learning policies based on the offline and generated data.

We study the sample complexity of learning one-hidden-layer convolutional neural networks (CNNs) with non-overlapping filters. We propose a novel algorithm called approximate gradient descent for training CNNs, and show that, with high probability, the proposed algorithm with random initialization grants a linear convergence to the ground-truth parameters up to statistical precision. Compared with existing work, our result applies to general non-trivial, monotonic and Lipschitz continuous activation functions including ReLU, Leaky ReLU, Sigmod and Softplus etc. Moreover, our sample complexity beats existing results in the dependency of the number of hidden nodes and filter size. In fact, our result matches the information-theoretic lower bound for learning one-hidden-layer CNNs with linear activation functions, suggesting that our sample complexity is tight. Our theoretical analysis is backed up by numerical experiments.

A major challenge in materials design is how to efficiently search the vast chemical design space to find the materials with desired properties. One effective strategy is to develop sampling algorithms that can exploit both explicit chemical knowledge and implicit composition rules embodied in the large materials database. Here, we propose a generative machine learning model (MatGAN) based on a generative adversarial network (GAN) for efficient generation of new hypothetical inorganic materials. Trained with materials from the ICSD database, our GAN model can generate hypothetical materials not existing in the training dataset, reaching a novelty of 92.53% when generating 2 million samples. The percentage of chemically valid (charge neutral and electronegativity balanced) samples out of all generated ones reaches 84.5% by our GAN when trained with materials from ICSD even though no such chemical rules are explicitly enforced in our GAN model, indicating its capability to learn implicit chemical composition rules. Our algorithm could be used to speed up inverse design or computational screening of inorganic materials.

This paper studies the prediction of chord progressions for jazz music by relying on machine learning models. The motivation of our study comes from the recent success of neural networks for performing automatic music composition. Although high accuracies are obtained in single-step prediction scenarios, most models fail to generate accurate multi-step chord predictions. In this paper, we postulate that this comes from the multi-scale structure of musical information and propose new architectures based on an iterative temporal aggregation of input labels. Specifically, the input and ground truth labels are merged into increasingly large temporal bags, on which we train a family of encoder-decoder networks for each temporal scale. In a second step, we use these pre-trained encoder bottleneck features at each scale in order to train a final encoder-decoder network. Furthermore, we rely on different reductions of the initial chord alphabet into three adapted chord alphabets. We perform evaluations against several state-of-the-art models and show that our multi-scale architecture outperforms existing methods in terms of accuracy and perplexity, while requiring relatively few parameters. We analyze musical properties of the results, showing the influence of downbeat position within the analysis window on accuracy, and evaluate errors using a musically-informed distance metric.

A BSTRACT In anomaly detection (AD), one seeks to identify whether a test sample is abnormal, given a data set of normal samples. A recent and promising approach to AD relies on deep generative models, such as variational autoencoders (V AEs), for unsupervised learning of the normal data distribution. In semi-supervised AD (SSAD), the data also includes a small sample of labeled anomalies. In this work, we propose two variational methods for training V AEs for SSAD. The intuitive idea in both methods is to train the encoder to'separate' between latent vectors for normal and outlier data. We show that this idea can be derived from principled probabilistic formulations of the problem, and propose simple and effective algorithms. Our methods can be applied to various data types, as we demonstrate on SSAD datasets ranging from natural images to astronomy and medicine, and can be combined with any V AE model architecture. When comparing to state-of-the-art SSAD methods that are not specific to particular data types, we obtain marked improvement in outlier detection. In its common formulation, training data is provided only for normal samples, while at test time, anomalous samples need to be detected. In the probabilistic AD approach, a model of the normal data distribution is learned, and the likelihood of a test sample under this model is thresholded for classification as normal or not. Recently, deep generative models such as variational autoencoders (V AEs, Kingma & Welling 2013) and generative adversarial networks (Goodfellow et al., 2014) have shown promise for learning data distributions in AD (An & Cho, 2015; Suh et al., 2016; Schlegl et al., 2017; Wang et al., 2017). Here, we consider the setting of semi-supervised AD (SSAD), where in addition to the normal samples, a small sample of labeled anomalies is provided (G ornitz et al., 2013). Most importantly, this set is too small to represent the range of possible anomalies, making classification methods (either supervised or semi-supervised) unsuitable. Instead, most approaches are based on'fixing' an unsupervised AD method to correctly classify the labeled anomalies, while still maintaining AD capabilities for unseen outliers (e.g., G ornitz et al., 2013; Mu noz-Mar ı et al., 2010; Ruff et al., 2019).

Automatic modulation classification (AMC) has been studied for more than a quarter of a century; however, it has been difficult to design a classifier that operates successfully under changing multipath fading conditions and other impairments. Recently, deep learning (DL)-based methods are adopted by AMC systems and major improvements are reported. In this paper, a novel convolutional neural network (CNN) classifier model is proposed to classify modulation classes in terms of their families, i.e., types. The proposed classifier is robust against realistic wireless channel impairments and in relation to that when the data sets that are utilized for testing and evaluating the proposed methods are considered, it is seen that RadioML2016.10a is the main dataset utilized for testing and evaluation of the proposed methods. However, the channel effects incorporated in this dataset and some others may lack the appropriate modeling of the real-world conditions since it only considers two distributions for channel models for a single tap configuration. Therefore, in this paper, a more comprehensive dataset, named as HisarMod2019.1, is also introduced, considering real-life applicability. HisarMod2019.1 includes 26 modulation classes passing through the channels with 5 different fading types and several numbers of taps for classification. It is shown that the proposed model performs better than the existing models in terms of both accuracy and training time under more realistic conditions. Even more, surpassed their performance when the RadioML2016.10a dataset is utilized.