Machine Learning
Unsupervised Data Augmentation (1904.12848v1)
Qizhe Xie, Zihang Dai, Eduard Hovy, Minh-Thang Luong, Quoc V. Le
2019-04-29
Despite its success, deep learning still needs large labeled datasets to succeed. Data augmentation has shown much promise in alleviating the need for more labeled data, but it so far has mostly been applied in supervised settings and achieved limited gains. In this work, we propose to apply data augmentation to unlabeled data in a semi-supervised learning setting. Our method, named Unsupervised Data Augmentation or UDA, encourages the model predictions to be consistent between an unlabeled example and an augmented unlabeled example. Unlike previous methods that use random noise such as Gaussian noise or dropout noise, UDA has a small twist in that it makes use of harder and more realistic noise generated by state-of-the-art data augmentation methods. This small twist leads to substantial improvements on six language tasks and three vision tasks even when the labeled set is extremely small. For example, on the IMDb text classification dataset, with only 20 labeled examples, UDA outperforms the state-of-the-art model trained on 25,000 labeled examples. On standard semi-supervised learning benchmarks, CIFAR-10 with 4,000 examples and SVHN with 1,000 examples, UDA outperforms all previous approaches and reduces more than
of the error rates of state-of-the-art methods: going from 7.66% to 5.27% and from 3.53% to 2.46% respectively. UDA also works well on datasets that have a lot of labeled data. For example, on ImageNet, with 1.3M extra unlabeled data, UDA improves the top-1/top-5 accuracy from 78.28/94.36% to 79.04/94.45% when compared to AutoAugment.
Optimal Sparse Decision Trees (1904.12847v1)
Xiyang Hu, Cynthia Rudin, Margo Seltzer
2019-04-29
Decision tree algorithms have been among the most popular algorithms for interpretable (transparent) machine learning since the early 1980's. The problem that has plagued decision tree algorithms since their inception is their lack of optimality, or lack of guarantees of closeness to optimality: decision tree algorithms are often greedy or myopic, and sometimes produce unquestionably suboptimal models. Hardness of decision tree optimization is both a theoretical and practical obstacle, and even careful mathematical programming approaches have not been able to solve these problems efficiently. This work introduces the first practical algorithm for optimal decision trees for binary variables. The algorithm is a co-design of analytical bounds that reduce the search space and modern systems techniques, including data structures and a custom bit-vector library. We highlight possible steps to improving the scalability and speed of future generations of this algorithm based on insights from our theory and experiments.
HIGAN: Cosmic Neutral Hydrogen with Generative Adversarial Networks (1904.12846v1)
Juan Zamudio-Fernandez, Atakan Okan, Francisco Villaescusa-Navarro, Seda Bilaloglu, Asena Derin Cengiz, Siyu He, Laurence Perreault Levasseur, Shirley Ho
2019-04-29
One of the most promising ways to observe the Universe is by detecting the 21cm emission from cosmic neutral hydrogen (HI) through radio-telescopes. Those observations can shed light on fundamental astrophysical questions only if accurate theoretical predictions are available. In order to maximize the scientific return of these surveys, those predictions need to include different observables and be precise on non-linear scales. Currently, one of the best ways to achieve this is via cosmological hydrodynamic simulations; however, the computational cost of these simulations is high -- tens of millions of CPU hours. In this work, we use Wasserstein Generative Adversarial Networks (WGANs) to generate new high-resolution (
) 3D realizations of cosmic HI at . We do so by sampling from a 100-dimension manifold, learned by the generator, that characterizes the fully non-linear abundance and clustering of cosmic HI from the state-of-the-art simulation IllustrisTNG. We show that different statistical properties of the produced samples -- 1D PDF, power spectrum, bispectrum, and void size function -- match very well those of IllustrisTNG, and outperform state-of-the-art models such as Halo Occupation Distributions (HODs). Our WGAN samples reproduce the abundance of HI across 9 orders of magnitude, from the Ly forest to Damped Lyman Absorbers. WGAN can produce new samples orders of magnitude faster than hydrodynamic simulations.
Adversarial Training for Free! (1904.12843v1)
Ali Shafahi, Mahyar Najibi, Amin Ghiasi, Zheng Xu, John Dickerson, Christoph Studer, Larry S. Davis, Gavin Taylor, Tom Goldstein
2019-04-29
Adversarial training, in which a network is trained on adversarial examples, is one of the few defenses against adversarial attacks that withstands strong attacks. Unfortunately, the high cost of generating strong adversarial examples makes standard adversarial training impractical on large-scale problems like ImageNet. We present an algorithm that eliminates the overhead cost of generating adversarial examples by recycling the gradient information computed when updating model parameters. Our "free" adversarial training algorithm achieves state-of-the-art robustness on CIFAR-10 and CIFAR-100 datasets at negligible additional cost compared to natural training, and can be 7 to 30 times faster than other strong adversarial training methods. Using a single workstation with 4 P100 GPUs and 2 days of runtime, we can train a robust model for the large-scale ImageNet classification task that maintains 40% accuracy against PGD attacks.
SEALion: a Framework for Neural Network Inference on Encrypted Data (1904.12840v1)
Tim van Elsloo, Giorgio Patrini, Hamish Ivey-Law
2019-04-29
We present SEALion: an extensible framework for privacy-preserving machine learning with homomorphic encryption. It allows one to learn deep neural networks that can be seamlessly utilized for prediction on encrypted data. The framework consists of two layers: the first is built upon TensorFlow and SEAL and exposes standard algebra and deep learning primitives; the second implements a Keras-like syntax for training and inference with neural networks. Given a required level of security, a user is abstracted from the details of the encoding and the encryption scheme, allowing quick prototyping. We present two applications that exemplifying the extensibility of our proposal, which are also of independent interest: i) improving efficiency of neural network inference by an activity sparsifier and ii) transfer learning by querying a server-side Variational AutoEncoder that can handle encrypted data.
The Step Decay Schedule: A Near Optimal, Geometrically Decaying Learning Rate Procedure (1904.12838v1)
Rong Ge, Sham M. Kakade, Rahul Kidambi, Praneeth Netrapalli
2019-04-29
There is a stark disparity between the step size schedules used in practical large scale machine learning and those that are considered optimal by the theory of stochastic approximation. In theory, most results utilize polynomially decaying learning rate schedules, while, in practice, the "Step Decay" schedule is among the most popular schedules, where the learning rate is cut every constant number of epochs (i.e. this is a geometrically decaying schedule). This work examines the step-decay schedule for the stochastic optimization problem of streaming least squares regression (both in the non-strongly convex and strongly convex case), where we show that a sharp theoretical characterization of an optimal learning rate schedule is far more nuanced than suggested by previous work. We focus specifically on the rate that is achievable when using the final iterate of stochastic gradient descent, as is commonly done in practice. Our main result provably shows that a properly tuned geometrically decaying learning rate schedule provides an exponential improvement (in terms of the condition number) over any polynomially decaying learning rate schedule. We also provide experimental support for wider applicability of these results, including for training modern deep neural networks.
Gated deep neural networks for implied volatility surfaces (1904.12834v1)
Yu Zheng, Yongxin Yang, Bowei Chen
2019-04-29
In this paper, we propose a gated deep neural network model to predict implied volatility surfaces. Conventional financial conditions and empirical evidence related to the implied volatility are incorporated into the neural network architecture design and calibration including no static arbitrage, boundaries, asymptotic slope and volatility smile. They are also satisfied empirically by the option data on the S&P 500 over a ten years period. Our proposed model outperforms the widely used surface stochastic volatility inspired model on the mean average percentage error in both in-sample and out-of-sample datasets. The research of this study has a fundamental methodological contribution to the emerging trend of applying the state-of-the-art information technology into business studies as our model provides a framework of integrating data-driven machine learning algorithms with financial theories and this framework can be easily extended and applied to solve other problems in finance or other business fields.
Wave Physics as an Analog Recurrent Neural Network (1904.12831v1)
Tyler W. Hughes, Ian A. D. Williamson, Momchil Minkov, Shanhui Fan
2019-04-29
Analog machine learning hardware platforms promise to be faster and more energy-efficient than their digital counterparts. Wave physics, as found in acoustics and optics, is a natural candidate for building analog processors for time-varying signals. Here we identify a mapping between the dynamics of wave physics, and the computation in recurrent neural networks. This mapping indicates that physical wave systems can be trained to learn complex features in temporal data, using standard training techniques for neural networks. As a demonstration, we show that an inversely-designed inhomogeneous medium can perform vowel classification on raw audio data by simple propagation of waves through such a medium, achieving performance that is comparable to a standard digital implementation of a recurrent neural network. These findings pave the way for a new class of analog machine learning platforms, capable of fast and efficient processing of information in its native domain.
Provably Correct Learning Algorithms in the Presence of Time-Varying Features Using a Variational Perspective (1903.04666v2)
Joseph E. Gaudio, Travis E. Gibson, Anuradha M. Annaswamy, Michael A. Bolender
2019-03-12
Features in machine learning problems are often time-varying and may be related to outputs in an algebraic or dynamical manner. The dynamic nature of these machine learning problems renders current accelerated gradient descent methods unstable or weakens their convergence guarantees. This paper proposes algorithms for the case when time-varying features are present, and demonstrates provable performance guarantees. We develop a variational perspective within a continuous time algorithm. This variational perspective includes, among other things, higher-order learning concepts and normalization, both of which stem from adaptive control, and allows stability to be established for dynamical machine learning problems. These higher-order algorithms are also examined for achieving accelerated learning in adaptive control. Simulations are provided to verify the theoretical results.
Bayesian Optimization for Policy Search via Online-Offline Experimentation (1904.01049v2)
Benjamin Letham, Eytan Bakshy
2019-04-01
Online field experiments are the gold-standard way of evaluating changes to real-world interactive machine learning systems. Yet our ability to explore complex, multi-dimensional policy spaces - such as those found in recommendation and ranking problems - is often constrained by the limited number of experiments that can be run simultaneously. To alleviate these constraints, we augment online experiments with an offline simulator and apply multi-task Bayesian optimization to tune live machine learning systems. We describe practical issues that arise in these types of applications, including biases that arise from using a simulator and assumptions for the multi-task kernel. We measure empirical learning curves which show substantial gains from including data from biased offline experiments, and show how these learning curves are consistent with theoretical results for multi-task Gaussian process generalization. We find that improved kernel inference is a significant driver of multi-task generalization. Finally, we show several examples of Bayesian optimization efficiently tuning a live machine learning system by combining offline and online experiments.