Today’s state-of-the-art machine learning models require massive labeled training sets–which usually do not exist for real-world applications. Instead, Snorkel is based around the new data programming paradigm, in which the developer focuses on writing a set of labeling functions, which are just scripts that programmatically label data. The resulting labels are noisy, but Snorkel automatically models this process – learning, essentially, which labeling functions are more accurate than others – and then uses this to train an end model (for example, a deep neural network in TensorFlow). …

In this paper, a novel approach, Inforence, is proposed to isolate the suspicious codes that likely contain faults. Inforence employs a feature selection method, based on mutual information, to identify those bug-related statements that may cause the program to fail. Because the majority of a program faults may be revealed as undesired joint effect of the program statements on each other and on program termination state, unlike the state-of-the-art methods, Inforence tries to identify and select groups of interdependent statements which altogether may affect the program failure. The interdependence amongst the statements is measured according to their mutual effect on each other and on the program termination state. To provide the context of failure, the selected bug-related statements are chained to each other, considering the program static structure. Eventually, the resultant cause-effect chains are ranked according to their combined causal effect on program failure. To validate Inforence, the results of our experiments with seven sets of programs include Siemens suite, gzip, grep, sed, space, make and bash are presented. The experimental results are then compared with those provided by different fault localization techniques for the both single-fault and multi-fault programs. The experimental results prove the outperformance of the proposed method compared to the state-of-the-art techniques. …

We present a novel approach to image manipulation and understanding by simultaneously learning to segment object masks, paste objects to another background image, and remove them from original images. For this purpose, we develop a novel generative model for compositional image generation, SEIGAN (Segment-Enhance-Inpaint Generative Adversarial Network), which learns these three operations together in an adversarial architecture with additional cycle consistency losses. To train, SEIGAN needs only bounding box supervision and does not require pairing or ground truth masks. SEIGAN produces better generated images (evaluated by human assessors) than other approaches and produces high-quality segmentation masks, improving over other adversarially trained approaches and getting closer to the results of fully supervised training. …

In this paper, a new approach for classification of target task using limited labeled target data as well as enormous unlabeled source data is proposed which is called self-taught learning. The target and source data can be drawn from different distributions. In the previous approaches, covariate shift assumption is considered where the marginal distributions p(x) change over domains and the conditional distributions p(y|x) remain the same. In our approach, we propose a new objective function which simultaneously learns a common space T(.) where the conditional distributions over domains p(T(x)|y) remain the same and learns robust SVM classifiers for target task using both source and target data in the new representation. Hence, in the proposed objective function, the hidden label of the source data is also incorporated. We applied the proposed approach on Caltech-256, MSRC+LMO datasets and compared the performance of our algorithm to the available competing methods. Our method has a superior performance to the successful existing algorithms. …

]]>

Reinforcement learning is a promising framework for solving control problems, but its use in practical situations is hampered by the fact that reward functions are often difficult to engineer. Specifying goals and tasks for autonomous machines, such as robots, is a significant challenge: conventionally, reward functions and goal states have been used to communicate objectives. But people can communicate objectives to each other simply by describing or demonstrating them. How can we build learning algorithms that will allow us to tell machines what we want them to do? In this work, we investigate the problem of grounding language commands as reward functions using inverse reinforcement learning, and argue that language-conditioned rewards are more transferable than language-conditioned policies to new environments. We propose language-conditioned reward learning (LC-RL), which grounds language commands as a reward function represented by a deep neural network. We demonstrate that our model learns rewards that transfer to novel tasks and environments on realistic, high-dimensional visual environments with natural language commands, whereas directly learning a language-conditioned policy leads to poor performance. …

On-line social networks, such as in Facebook and Twitter, are often studied from the perspective of friendship ties between agents in the network. Adversarial ties, however, also play an important role in the structure and function of social networks, but are often hidden. Underlying generative mechanisms of social networks are predicted by structural balance theory, which postulates that triads of agents, prefer to be transitive, where friends of friends are more likely friends, or anti-transitive, where adversaries of adversaries become friends. The previously proposed Iterated Local Transitivity (ILT) and Iterated Local Anti-Transitivity (ILAT) models incorporated transitivity and anti-transitivity, respectively, as evolutionary mechanisms. These models resulted in graphs with many observable properties of social networks, such as low diameter, high clustering, and densification. We propose a new, generative model, referred to as the Iterated Local Model (ILM) for social networks synthesizing both transitive and anti-transitive triads over time. In ILM, we are given a countably infinite binary sequence as input, and that sequence determines whether we apply a transitive or an anti-transitive step. The resulting model exhibits many properties of complex networks observed in the ILT and ILAT models. In particular, for any input binary sequence, we show that asymptotically the model generates finite graphs that densify, have clustering coefficient bounded away from 0, have diameter at most 3, and exhibit bad spectral expansion. We also give a thorough analysis of the chromatic number, domination number, Hamiltonicity, and isomorphism types of induced subgraphs of ILM graphs. …

Long Short-Term Memory (LSTM) Recurrent Neural networks (RNNs) rely on gating signals, each driven by a function of a weighted sum of at least 3 components: (i) one of an adaptive weight matrix multiplied by the incoming external input vector sequence, (ii) one adaptive weight matrix multiplied by the previous memory/state vector, and (iii) one adaptive bias vector. In effect, they augment the simple Recurrent Neural Networks (sRNNs) structure with the addition of a ‘memory cell’ and the incorporation of at most 3 gating signals. The standard LSTM structure and components encompass redundancy and overly increased parameterization. In this paper, we systemically introduce variants of the LSTM RNNs, referred to as SLIM LSTMs. These variants express aggressively reduced parameterizations to achieve computational saving and/or speedup in (training) performance—while necessarily retaining (validation accuracy) performance comparable to the standard LSTM RNN. …

Neural networks designed for the task of classification have become a commodity in recent years. Many works target the development of better networks, which results in a complexification of their architectures with more layers, multiple sub-networks, or even the combination of multiple classifiers. In this paper, we show how to redesign a simple network to reach excellent performances, which are better than the results reproduced with CapsNet on several datasets, by replacing a layer with a Hit-or-Miss layer. This layer contains activated vectors, called capsules, that we train to hit or miss a central capsule by tailoring a specific centripetal loss function. We also show how our network, named HitNet, is capable of synthesizing a representative sample of the images of a given class by including a reconstruction network. This possibility allows to develop a data augmentation step combining information from the data space and the feature space, resulting in a hybrid data augmentation process. In addition, we introduce the possibility for HitNet, to adopt an alternative to the true target when needed by using the new concept of ghost capsules, which is used here to detect potentially mislabeled images in the training data. …

]]>

Data movement is a major bottleneck in parallel data-intensive applications. In response to this problem, researchers have proposed new runtimes and intermediate representations (IRs) that apply optimizations such as loop fusion under existing library APIs. Even though these runtimes generally do no require changes to user code, they require intrusive changes to the library itself: often, all the library functions need to be rewritten for a new IR or virtual machine. In this paper, we propose a new abstraction called splitability annotations (SAs) that enables key data movement optimizations on black-box library functions. SAs only require that users add an annotation for existing, unmodified functions and implement a small API to split data values in the library. Together, this interface describes how to partition values that are passed among functions to enable data pipelining and automatic parallelization while respecting each library’s correctness constraints. We implement SAs in a system called Mozart. Without modifying any library function, on workloads using NumPy and Pandas in Python and Intel MKL in C, Mozart provides performance competitive with intrusive solutions that require rewriting libraries in many cases, can sometimes improve performance over past systems by up to 2x, and accelerates workloads by up to 30x. …

Methodology: Remove one observation. Training the rest of data that are sampled without replacement and given this observation’s input, predict the response back. Replicate this N times and for each response, take a sample from these replicates with replacement. Average each responses of sample and again replicate this step N time for each observation. Approximate these N new responses and generate another N responses y’. Training these y’ and predict to have N responses of each testing observation. The average of N is the final prediction. Each observation will do the same. …

High-dimensional prediction models are increasingly used to analyze biological data such as neuroimaging of genetic data sets. However, classical penalized algorithms yield to dense solutions that are difficult to interpret without arbitrary thresholding. Alternatives based on sparsity-inducing penalties suffer from coefficient instability. Complex structured sparsity-inducing penalties are a promising approach to force the solution to adhere to some domain-specific constraints and thus offering new perspectives in biomarker identification. We propose a generic optimization framework that can combine any smooth convex loss function with: (i) penalties whose proximal operator is known and (ii) with a large range of complex, non-smooth convex structured penalties such as total variation, or overlapping group lasso. Although many papers have addressed a similar goal, few have tackled it in such a generic way and in the context of high-dimensional data. The proposed continuation algorithm, called \textit{CONESTA}, dynamically smooths the complex penalties to avoid the computation of proximal operators, that are either not known or expensive to compute. The decreasing sequence of smoothing parameters is dynamically adapted, using the duality gap, in order to maintain the optimal convergence speed towards any globally desired precision with duality gap guarantee. First, we demonstrate, on both simulated data and on experimental MRI data, that CONESTA outperforms the excessive gap method, ADMM, proximal gradient smoothing (without continuation) and inexact FISTA in terms of convergence speed and/or precision of the solution. Second, on the experimental MRI data set, we establish the superiority of structured sparsity-inducing penalties ($\ell_1$ and total variation) over non-structured methods in terms of the recovery of meaningful and stable groups of predictive variables. …

We propose a novel reinforcement learning algorithm that approximates solutions to the problem of discounted optimal stopping in an irreducible, uniformly ergodic Markov chain evolving on a compact subset of $\mathbb R^n$. A dynamic programming approach has been taken by Tsitsikilis and Van Roy to solve this problem, wherein they propose a Q-learning algorithm to estimate the value function, in a linear function approximation setting. The Zap-Q learning algorithm proposed in this work is the first algorithm that is designed to achieve {optimal asymptotic variance}. We prove convergence of the algorithm using ODE analysis, and the optimal asymptotic variance property is reflected via fast convergence in a finance example. …

]]>

In this study, we introduce EdgeSegNet, a compact deep convolutional neural network for the task of semantic segmentation. A human-machine collaborative design strategy is leveraged to create EdgeSegNet, where principled network design prototyping is coupled with machine-driven design exploration to create networks with customized module-level macroarchitecture and microarchitecture designs tailored for the task. Experimental results showed that EdgeSegNet can achieve semantic segmentation accuracy comparable with much larger and computationally complex networks (>20x} smaller model size than RefineNet) as well as achieving an inference speed of ~38.5 FPS on an NVidia Jetson AGX Xavier. As such, the proposed EdgeSegNet is well-suited for low-power edge scenarios. …

Most existing event extraction (EE) methods merely extract event arguments within the sentence scope. However, such sentence-level EE methods struggle to handle soaring amounts of documents from emerging applications, such as finance, legislation, health, etc., where event arguments always scatter across different sentences, and even multiple such event mentions frequently co-exist in the same document. To address these challenges, we propose a novel end-to-end solution, Doc2EDAG, which can efficiently generate an entity-based directed acyclic graph to fulfill the document-level EE (DEE). Moreover, we reformalize a DEE task with the no-trigger-words design to ease the document-level event labeling. To demonstrate the effectiveness of Doc2EDAG, we build a large-scale real-world dataset consisting of Chinese financial announcements with the challenges mentioned above. Extensive experiments with comprehensive analyses illustrate the superiority of Doc2EDAG over state-of-the-art methods. …

Image classification is an important task in today’s world with many applications from socio-technical to safety-critical domains. The recent advent of Deep Neural Network (DNN) is the key behind such a wide-spread success. However, such wide adoption comes with the concerns about the reliability of these systems, as several erroneous behaviors have already been reported in many sensitive and critical circumstances. Thus, it has become crucial to rigorously test the image classifiers to ensure high reliability. Many reported erroneous cases in popular neural image classifiers appear because the models often confuse one class with another, or show biases towards some classes over others. These errors usually violate some group properties. Most existing DNN testing and verification techniques focus on per image violations and thus fail to detect such group-level confusions or biases. In this paper, we design, implement and evaluate DeepInspect, a white box testing tool, for automatically detecting confusion and bias of DNN-driven image classification applications. We evaluate DeepInspect using popular DNN-based image classifiers and detect hundreds of classification mistakes. Some of these cases are able to expose potential biases of the network towards certain populations. DeepInspect further reports many classification errors in state-of-the-art robust models. …

In the theory of probability for stochastic processes, the reflection principle for a Wiener process states that if the path of a Wiener process f(t) reaches a value f(s) = a at time t = s, then the subsequent path after time s has the same distribution as the reflection of the subsequent path about the value a. More formally, the reflection principle refers to a lemma concerning the distribution of the supremum of the Wiener process, or Brownian motion. The result relates the distribution of the supremum of Brownian motion up to time t to the distribution of the process at time t. It is a corollary of the strong Markov property of Brownian motion.

A Direct Proof of the Reflection Principle for Brownian Motion …

]]>

The emotion cause extraction (ECE) task aims at discovering the potential causes behind a certain emotion expression in a document. Techniques including rule-based methods, traditional machine learning methods and deep neural networks have been proposed to solve this task. However, most of the previous work considered ECE as a set of independent clause classification problems and ignored the relations between multiple clauses in a document. …

Many natural language processing tasks solely rely on sparse dependencies between a few tokens in a sentence. Soft attention mechanisms show promising performance in modeling local/global dependencies by soft probabilities between every two tokens, but they are not effective and efficient when applied to long sentences. By contrast, hard attention mechanisms directly select a subset of tokens but are difficult and inefficient to train due to their combinatorial nature. In this paper, we integrate both soft and hard attention into one context fusion model, ‘reinforced self-attention (ReSA)’, for the mutual benefit of each other. In ReSA, a hard attention trims a sequence for a soft self-attention to process, while the soft attention feeds reward signals back to facilitate the training of the hard one. For this purpose, we develop a novel hard attention called ‘reinforced sequence sampling (RSS)’, selecting tokens in parallel and trained via policy gradient. Using two RSS modules, ReSA efficiently extracts the sparse dependencies between each pair of selected tokens. We finally propose an RNN/CNN-free sentence-encoding model, ‘reinforced self-attention network (ReSAN)’, solely based on ReSA. It achieves state-of-the-art performance on both Stanford Natural Language Inference (SNLI) and Sentences Involving Compositional Knowledge (SICK) datasets. …

In many applications requiring multiple inputs to obtain a desired output, if any of the input data is missing, it often introduces large amounts of bias. Although many techniques have been developed for imputing missing data, the image imputation is still difficult due to complicated nature of natural images. To address this problem, here we proposed a novel framework for missing image data imputation, called Collaborative Generative Adversarial Network (CollaGAN). CollaGAN converts an image imputation problem to a multi-domain images-to-image translation task so that a single generator and discriminator network can successfully estimate the missing data using the remaining clean data set. We demonstrate that CollaGAN produces the images with a higher visual quality compared to the existing competing approaches in various image imputation tasks. …

To facilitate efficient embedded and hardware implementations of deep neural networks (DNNs), two important categories of DNN model compression techniques: weight pruning and weight quantization are investigated. The former leverages the redundancy in the number of weights, whereas the latter leverages the redundancy in bit representation of weights. However, there lacks a systematic framework of joint weight pruning and quantization of DNNs, thereby limiting the available model compression ratio. Moreover, the computation reduction, energy efficiency improvement, and hardware performance overhead need to be accounted for besides simply model size reduction. To address these limitations, we present ADMM-NN, the first algorithm-hardware co-optimization framework of DNNs using Alternating Direction Method of Multipliers (ADMM), a powerful technique to deal with non-convex optimization problems with possibly combinatorial constraints. The first part of ADMM-NN is a systematic, joint framework of DNN weight pruning and quantization using ADMM. It can be understood as a smart regularization technique with regularization target dynamically updated in each ADMM iteration, thereby resulting in higher performance in model compression than prior work. The second part is hardware-aware DNN optimizations to facilitate hardware-level implementations. Without accuracy loss, we can achieve 85$\times$ and 24$\times$ pruning on LeNet-5 and AlexNet models, respectively, significantly higher than prior work. The improvement becomes more significant when focusing on computation reductions. Combining weight pruning and quantization, we achieve 1,910$\times$ and 231$\times$ reductions in overall model size on these two benchmarks, when focusing on data storage. Highly promising results are also observed on other representative DNNs such as VGGNet and ResNet-50. …

]]>