Polygon-RNN++
Manually labeling datasets with object masks is extremely time consuming. In this work, we follow the idea of Polygon-RNN to produce polygonal annotations of objects interactively using humans-in-the-loop. We introduce several important improvements to the model: 1) we design a new CNN encoder architecture, 2) show how to effectively train the model with Reinforcement Learning, and 3) significantly increase the output resolution using a Graph Neural Network, allowing the model to accurately annotate high-resolution objects in images. Extensive evaluation on the Cityscapes dataset shows that our model, which we refer to as Polygon-RNN++, significantly outperforms the original model in both automatic (10% absolute and 16% relative improvement in mean IoU) and interactive modes (requiring 50% fewer clicks by annotators). We further analyze the cross-domain scenario in which our model is trained on one dataset, and used out of the box on datasets from varying domains. The results show that Polygon-RNN++ exhibits powerful generalization capabilities, achieving significant improvements over existing pixel-wise methods. Using simple online fine-tuning we further achieve a high reduction in annotation time for new datasets, moving a step closer towards an interactive annotation tool to be used in practice. …

Optimal Sparse Decision Tree
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. …

Population-Attributable Fraction (PAF)
The contribution of a risk factor to a disease or a death is quantified using the population attributable fraction (PAF). PAF is the proportional reduction in population disease or mortality that would occur if exposure to a risk factor were reduced to an alternative ideal exposure scenario (eg. no tobacco use). Many diseases are caused by multiple risk factors, and individual risk factors may interact in their impact on overall risk of disease. As a result, PAFs for individual risk factors often overlap and add up to more than 100 percent.
Causal inference with multi-state models – estimands and estimators of the population-attributable fraction …

Nature Language Inference (NLI)
Nature language inference (NLI) task is a predictive task of determining the inference relationship of a pair of natural language sentences. With the increasing popularity of NLI, many state-of-the-art predictive models have been proposed with impressive performances. However, several works have noticed the statistical irregularities in the collected NLI data set that may result in an over-estimated performance of these models and proposed remedies. …

Autoencoding Binary Classifier (ABC)
We propose the Autoencoding Binary Classifiers (ABC), a novel supervised anomaly detector based on the Autoencoder (AE). There are two main approaches in anomaly detection: supervised and unsupervised. The supervised approach accurately detects the known anomalies included in training data, but it cannot detect the unknown anomalies. Meanwhile, the unsupervised approach can detect both known and unknown anomalies that are located away from normal data points. However, it does not detect known anomalies as accurately as the supervised approach. Furthermore, even if we have labeled normal data points and anomalies, the unsupervised approach cannot utilize these labels. The ABC is a probabilistic binary classifier that effectively exploits the label information, where normal data points are modeled using the AE as a component. By maximizing the likelihood, the AE in the proposed ABC is trained to minimize the reconstruction error for normal data points, and to maximize it for known anomalies. Since our approach becomes able to reconstruct the normal data points accurately and fails to reconstruct the known and unknown anomalies, it can accurately discriminate both known and unknown anomalies from normal data points. Experimental results show that the ABC achieves higher detection performance than existing supervised and unsupervised methods. …

Parikh Matrix
Parikh Matrices are a newly developed tool for studying numerical properties of words in terms of their (scattered) subwords. They were introduced by Mateescu et al. in 2000 and continuously received attention from the research community ever since.
Mateescu et al (2000) introduced an interesting new tool, called Parikh matrix, to study in terms of subwords, the numerical properties of words over an alphabet. The Parikh matrix gives more information than the well-known Parikh vector of a word which counts only occurrences of symbols in a word. …

Graph Node-Feature Convolution
Graph convolutional network (GCN) is an emerging neural network approach. It learns new representation of a node by aggregating feature vectors of all neighbors in the aggregation process without considering whether the neighbors or features are useful or not. Recent methods have improved solutions by sampling a fixed size set of neighbors, or assigning different weights to different neighbors in the aggregation process, but features within a feature vector are still treated equally in the aggregation process. In this paper, we introduce a new convolution operation on regular size feature maps constructed from features of a fixed node bandwidth via sampling to get the first-level node representation, which is then passed to a standard GCN to learn the second-level node representation. Experiments show that our method outperforms competing methods in semi-supervised node classification tasks. Furthermore, our method opens new doors for exploring new GCN architectures, particularly deeper GCN models. …

Higher-Order Kolmogorov-Smirnov Test
We present an extension of the Kolmogorov-Smirnov (KS) two-sample test, which can be more sensitive to differences in the tails. Our test statistic is an integral probability metric (IPM) defined over a higher-order total variation ball, recovering the original KS test as its simplest case. We give an exact representer result for our IPM, which generalizes the fact that the original KS test statistic can be expressed in equivalent variational and CDF forms. For small enough orders ($k \leq 5$), we develop a linear-time algorithm for computing our higher-order KS test statistic; for all others ($k \geq 6$), we give a nearly linear-time approximation. We derive the asymptotic null distribution for our test, and show that our nearly linear-time approximation shares the same asymptotic null. Lastly, we complement our theory with numerical studies. …

Progressive Operational Perceptron With Memory
Generalized Operational Perceptron (GOP) was proposed to generalize the linear neuron model in the traditional Multilayer Perceptron (MLP) and this model can mimic the synaptic connections of the biological neurons that have nonlinear neurochemical behaviours. Progressive Operational Perceptron (POP) is a multilayer network composing of GOPs which is formed layer-wise progressively. In this work, we propose major modifications that can accelerate as well as augment the progressive learning procedure of POP by incorporating an information-preserving, linear projection path from the input to the output layer at each progressive step. The proposed extensions can be interpreted as a mechanism that provides direct information extracted from the previously learned layers to the network, hence the term ‘memory’. This allows the network to learn deeper architectures with better data representations. An extensive set of experiments show that the proposed modifications can surpass the learning capability of the original POPs and other related algorithms. …

X-Armed Bandits
We propose and analyze StoROO, an algorithm for risk optimization on stochastic black-box functions derived from StoOO. Motivated by risk-averse decision making fields like agriculture, medicine, biology or finance, we do not focus on the mean payoff but on generic functionals of the return distribution, like for example quantiles. We provide a generic regret analysis of StoROO. Inspired by the bandit literature and black-box mean optimizers, StoROO relies on the possibility to construct confidence intervals for the targeted functional based on random-size samples. We explain in detail how to construct them for quantiles, providing tight bounds based on Kullback-Leibler divergence. The interest of these tight bounds is highlighted by numerical experiments that show a dramatic improvement over standard approaches. …

Event Store
An event store is a type of database optimized for storage of events. Conceptually, in an event store, only the events of a dossier or policy are stored. The idea behind it is that the dossier or policy can be derived from these events. The events (and their corresponding data) are the only ‘real’ facts that should be stored in the database. The instantiation of all other objects can be derived from these events. The code instantiates these objects in memory. In an event store database, this means that all objects that should be instantiated, are not stored in the database. Instead these objects are instantiated ‘on the fly’ in memory by the code based on the events. After usage of these objects (e.g. shown in a user interface), the instantiated objects are removed from memory. For example, the event store concept of a database can be applied to insurance policies or pension dossiers. In these policies or dossiers the instantiation of each object that make up the dossier or policy (the person, partner(s), employments, etc.) can be derived and can be instantiated in memory based on the real world events. …

Difference Compression
Optimizing distributed learning systems is an art of balancing between computation and communication. There have been two lines of research that try to deal with slower networks: {\em quantization} for low bandwidth networks, and {\em decentralization} for high latency networks. In this paper, we explore a natural question: {\em can the combination of both decentralization and quantization lead to a system that is robust to both bandwidth and latency?} Although the system implication of such combination is trivial, the underlying theoretical principle and algorithm design is challenging: simply quantizing data sent in a decentralized training algorithm would accumulate the error. In this paper, we develop a framework of quantized, decentralized training and propose two different strategies, which we call {\em extrapolation compression} and {\em difference compression}. We analyze both algorithms and prove both converge at the rate of $O(1/\sqrt{nT})$ where $n$ is the number of workers and $T$ is the number of iterations, matching the {\rc convergence} rate for full precision, centralized training. We evaluate our algorithms on training deep learning models, and find that our proposed algorithm outperforms the best of merely decentralized and merely quantized algorithm significantly for networks with {\em both} high latency and low bandwidth. …

Polypus
In this paper we propose a new parallel architecture based on Big Data technologies for real-time sentiment analysis on microblogging posts. Polypus is a modular framework that provides the following functionalities: (1) massive text extraction from Twitter, (2) distributed non-relational storage optimized for time range queries, (3) memory-based intermodule buffering, (4) real-time sentiment classification, (5) near real-time keyword sentiment aggregation in time series, (6) a HTTP API to interact with the Polypus cluster and (7) a web interface to analyze results visually. The whole architecture is self-deployable and based on Docker containers. …

Real-Time Anomaly Detection System (RADS)
Cybersecurity attacks in Cloud data centres are increasing alongside the growth of the Cloud services market. Existing research proposes a number of anomaly detection systems for detecting such attacks. However, these systems encounter a number of challenges, specifically due to the unknown behaviour of the attacks and the occurrence of genuine Cloud workload spikes, which must be distinguished from attacks. In this paper, we discuss these challenges and investigate the issues with the existing Cloud anomaly detection approaches. Then, we propose a Real-time Anomaly Detection System (RADS) for Cloud data centres, which uses a one class classification algorithm and a window-based time series analysis to address the challenges. Specifically, RADS can detect VM-level anomalies occurring due to DDoS and cryptomining attacks. We evaluate the performance of RADS by running lab-based experiments and by using real-world Cloud workload traces. Evaluation results demonstrate that RADS can achieve 90-95% accuracy with a low false positive rate of 0-3%. The results further reveal that RADS experiences fewer false positives when using its window-based time series analysis in comparison to using state-of-the-art average or entropy based analysis. …

Sparsely Connected Convolutional Network (SparseNet)
Residual learning with skip connections permits training ultra-deep neural networks and obtains superb performance. Building in this direction, DenseNets proposed a dense connection structure where each layer is directly connected to all of its predecessors. The densely connected structure leads to better information flow and feature reuse. However, the overly dense skip connections also bring about the problems of potential risk of overfitting, parameter redundancy and large memory consumption. In this work, we analyze the feature aggregation patterns of ResNets and DenseNets under a uniform aggregation view framework. We show that both structures densely gather features from previous layers in the network but combine them in their respective ways: summation (ResNets) or concatenation (DenseNets). We compare the strengths and drawbacks of these two aggregation methods and analyze their potential effects on the networks’ performance. Based on our analysis, we propose a new structure named SparseNets which achieves better performance with fewer parameters than DenseNets and ResNets. …

Shewhart Control Chart
Control charts, also known as Shewhart charts (after Walter A. Shewhart) or process-behavior charts, are a statistical process control tool used to determine if a manufacturing or business process is in a state of control. If analysis of the control chart indicates that the process is currently under control (i.e., is stable, with variation only coming from sources common to the process), then no corrections or changes to process control parameters are needed or desired. In addition, data from the process can be used to predict the future performance of the process. If the chart indicates that the monitored process is not in control, analysis of the chart can help determine the sources of variation, as this will result in degraded process performance. A process that is stable but operating outside desired (specification) limits (e.g., scrap rates may be in statistical control but above desired limits) needs to be improved through a deliberate effort to understand the causes of current performance and fundamentally improve the process. The control chart is one of the seven basic tools of quality control. Typically control charts are used for time-series data, though they can be used for data that have logical comparability (i.e. you want to compare samples that were taken all at the same time, or the performance of different individuals); however the type of chart used to do this requires consideration. …

V-Monotone Independence
We introduce and study a new notion of non-commutative independence, called V-monotone independence, which can be viewed as an extension of the monotone independence of Muraki. We investigate the combinatorics of mixed moments of V-monotone random variables and prove the central limit theorem. We obtain a combinatorial formula for the limit moments and we find the solution of the differential equation for the moment generating function in the implicit form. …

Feature Pyramid Hashing
In recent years, deep-networks-based hashing has become a leading approach for large-scale image retrieval. Most deep hashing approaches use the high layer to extract the powerful semantic representations. However, these methods have limited ability for fine-grained image retrieval because the semantic features extracted from the high layer are difficult in capturing the subtle differences. To this end, we propose a novel two-pyramid hashing architecture to learn both the semantic information and the subtle appearance details for fine-grained image search. Inspired by the feature pyramids of convolutional neural network, a vertical pyramid is proposed to capture the high-layer features and a horizontal pyramid combines multiple low-layer features with structural information to capture the subtle differences. To fuse the low-level features, a novel combination strategy, called consensus fusion, is proposed to capture all subtle information from several low-layers for finer retrieval. Extensive evaluation on two fine-grained datasets CUB-200-2011 and Stanford Dogs demonstrate that the proposed method achieves significant performance compared with the state-of-art baselines. …

Empirical Bayes Method
Empirical Bayes methods are procedures for statistical inference in which the prior distribution is estimated from the data. This approach stands in contrast to standard Bayesian methods, for which the prior distribution is fixed before any data are observed. Despite this difference in perspective, empirical Bayes may be viewed as an approximation to a fully Bayesian treatment of a hierarchical model wherein the parameters at the highest level of the hierarchy are set to their most likely values, instead of being integrated out. Empirical Bayes, also known as maximum marginal likelihood, represents one approach for setting hyperparameters. …

Deep Abstract Q-Network
We examine the problem of learning and planning on high-dimensional domains with long horizons and sparse rewards. Recent approaches have shown great successes in many Atari 2600 domains. However, domains with long horizons and sparse rewards, such as Montezuma’s Revenge and Venture, remain challenging for existing methods. Methods using abstraction (Dietterich 2000; Sutton, Precup, and Singh 1999) have shown to be useful in tackling long-horizon problems. We combine recent techniques of deep reinforcement learning with existing model-based approaches using an expert-provided state abstraction. We construct toy domains that elucidate the problem of long horizons, sparse rewards and high-dimensional inputs, and show that our algorithm significantly outperforms previous methods on these domains. Our abstraction-based approach outperforms Deep Q-Networks (Mnih et al. 2015) on Montezuma’s Revenge and Venture, and exhibits backtracking behavior that is absent from previous methods. …

Tiny Deeply Supervised Object Detection (Tiny-DSOD)
Object detection has made great progress in the past few years along with the development of deep learning. However, most current object detection methods are resource hungry, which hinders their wide deployment to many resource restricted usages such as usages on always-on devices, battery-powered low-end devices, etc. This paper considers the resource and accuracy trade-off for resource-restricted usages during designing the whole object detection framework. Based on the deeply supervised object detection (DSOD) framework, we propose Tiny-DSOD dedicating to resource-restricted usages. Tiny-DSOD introduces two innovative and ultra-efficient architecture blocks: depthwise dense block (DDB) based backbone and depthwise feature-pyramid-network (D-FPN) based front-end. We conduct extensive experiments on three famous benchmarks (PASCAL VOC 2007, KITTI, and COCO), and compare Tiny-DSOD to the state-of-the-art ultra-efficient object detection solutions such as Tiny-YOLO, MobileNet-SSD (v1 & v2), SqueezeDet, Pelee, etc. Results show that Tiny-DSOD outperforms these solutions in all the three metrics (parameter-size, FLOPs, accuracy) in each comparison. For instance, Tiny-DSOD achieves 72.1% mAP with only 0.95M parameters and 1.06B FLOPs, which is by far the state-of-the-arts result with such a low resource requirement. …

Asymmetric Deep Semantic Quantization (ADSQ)
Due to its fast retrieval and storage efficiency capabilities, hashing has been widely used in nearest neighbor retrieval tasks. By using deep learning based techniques, hashing can outperform non-learning based hashing in many applications. However, there are some limitations to previous learning based hashing methods (e.g., the learned hash codes are not discriminative due to the hashing methods being unable to discover rich semantic information and the training strategy having difficulty optimizing the discrete binary codes). In this paper, we propose a novel learning based hashing method, named \textbf{\underline{A}}symmetric \textbf{\underline{D}}eep \textbf{\underline{S}}emantic \textbf{\underline{Q}}uantization (\textbf{ADSQ}). \textbf{ADSQ} is implemented using three stream frameworks, which consists of one \emph{LabelNet} and two \emph{ImgNets}. The \emph{LabelNet} leverages three fully-connected layers, which is used to capture rich semantic information between image pairs. For the two \emph{ImgNets}, they each adopt the same convolutional neural network structure, but with different weights (i.e., asymmetric convolutional neural networks). The two \emph{ImgNets} are used to generate discriminative compact hash codes. Specifically, the function of the \emph{LabelNet} is to capture rich semantic information that is used to guide the two \emph{ImgNets} in minimizing the gap between the real-continuous features and discrete binary codes. By doing this, \textbf{ADSQ} can make full use of the most critical semantic information to guide the feature learning process and consider the consistency of the common semantic space and Hamming space. Results from our experiments demonstrate that \textbf{ADSQ} can generate high discriminative compact hash codes and it outperforms current state-of-the-art methods on three benchmark datasets, CIFAR-10, NUS-WIDE, and ImageNet. …

Deterministic Parallel Analysis (DPA)
Factor analysis is widely used in many application areas. The first step, choosing the number of factors, remains a serious challenge. One of the most popular methods is parallel analysis (PA), which compares the observed factor strengths to simulated ones under a noise-only model. % Abstracts are commonly just one paragraph. This paper presents a deterministic version of PA (DPA), which is faster and more reproducible than PA. We show that DPA selects large factors and does not select small factors just like [Dobriban, 2017] shows for PA. Both PA and DPA are prone to a shadowing phenomenon in which a strong factor makes it hard to detect smaller but more interesting factors. We develop a deflated version of DPA (DDPA) that counters shadowing. By raising the decision threshold in DDPA, a new method (DDPA+) also improves estimation accuracy. We illustrate our methods on data from the Human Genome Diversity Project (HGDP). There PA and DPA select seemingly too many factors, while DDPA+ selects only a few. A Matlab implementation is available. …

Feature Importance and Bilinear feature Interaction NETwork (FiBiNET)
Advertising and feed ranking are essential to many Internet companies such as Facebook and Sina Weibo. Among many real-world advertising and feed ranking systems, click through rate (CTR) prediction plays a central role. There are many proposed models in this field such as logistic regression, tree based models, factorization machine based models and deep learning based CTR models. However, many current works calculate the feature interactions in a simple way such as Hadamard product and inner product and they care less about the importance of features. In this paper, a new model named FiBiNET as an abbreviation for Feature Importance and Bilinear feature Interaction NETwork is proposed to dynamically learn the feature importance and fine-grained feature interactions. On the one hand, the FiBiNET can dynamically learn the importance of features via the Squeeze-Excitation network (SENET) mechanism; on the other hand, it is able to effectively learn the feature interactions via bilinear function. We conduct extensive experiments on two real-world datasets and show that our shallow model outperforms other shallow models such as factorization machine(FM) and field-aware factorization machine(FFM). In order to improve performance further, we combine a classical deep neural network(DNN) component with the shallow model to be a deep model. The deep FiBiNET consistently outperforms the other state-of-the-art deep models such as DeepFM and extreme deep factorization machine(XdeepFM). …

Multiple Policy Value Monte Carlo Tree Search (MPV-MCTS)
Many of the strongest game playing programs use a combination of Monte Carlo tree search (MCTS) and deep neural networks (DNN), where the DNNs are used as policy or value evaluators. Given a limited budget, such as online playing or during the self-play phase of AlphaZero (AZ) training, a balance needs to be reached between accurate state estimation and more MCTS simulations, both of which are critical for a strong game playing agent. Typically, larger DNNs are better at generalization and accurate evaluation, while smaller DNNs are less costly, and therefore can lead to more MCTS simulations and bigger search trees with the same budget. This paper introduces a new method called the multiple policy value Monte Carlo tree search (MPV-MCTS), which combines multiple policy value neural networks (PV-NNs) of various sizes to retain advantages of each network, where two PV-NNs f_S and f_L are used in this paper. We show through experiments on the game NoGo that a combined f_S and f_L MPV-MCTS outperforms single PV-NN with policy value MCTS, called PV-MCTS. Additionally, MPV-MCTS also outperforms PV-MCTS for AZ training. …

Educational Data Mining (EDM)
Educational Data Mining (EDM) describes a research field concerned with the application of data mining, machine learning and statistics to information generated from educational settings (e.g., universities and intelligent tutoring systems). At a high level, the field seeks to develop and improve methods for exploring this data, which often has multiple levels of meaningful hierarchy, in order to discover new insights about how people learn in the context of such settings. In doing so, EDM has contributed to theories of learning investigated by researchers in educational psychology and the learning sciences. The field is closely tied to that of learning analytics, and the two have been compared and contrasted. …

Surprise-Based Learning
This paper presents a learning algorithm known as surprise-based learning (SBL) capable of providing a physical robot the ability to autonomously learn and plan in an unknown environment without any prior knowledge of its actions or their impact on the environment. This is achieved by creating a model of the environment using prediction rules. A prediction rule describes the observations of the environment prior to the execution of an action and the forecasted or predicted observation of the environment after the action. The algorithm learns by investigating ‘surprises’, which are inconsistencies between the predictions and observed outcome. SBL has been successfully demonstrated on a modular robot learning and navigating in a small static environment.
Surprise-based learning is capable of providing a physical robot the ability to autonomously learn and plan in an unknown environment without any prior knowledge of its actions or their impact on the environment. This is achieved by creating a model of the environment using prediction rules. A prediction rule describes the observations of the environment prior to the execution of an action and the forecasted or predicted observation of the environment after the action. The algorithm learns by investigating ‘surprises’, which are inconsistencies between the predictions and observed outcome. SBL has been successfully demonstrated on a modular robot learning and navigating in an office-environment, and other real-world applications below..
An Approximate Bayesian Approach to Surprise-Based Learning
Surprise-Based Learning (SBL) …

Few-Shot Self Reminder (FSR)
Deep neural networks are known to suffer the catastrophic forgetting problem, where they tend to forget the knowledge from the previous tasks when sequentially learning new tasks. Such failure hinders the application of deep learning based vision system in continual learning settings. In this work, we present a simple yet surprisingly effective way of preventing catastrophic forgetting. Our method, called Few-shot Self Reminder (FSR), regularizes the neural net from changing its learned behaviour by performing logit matching on selected samples kept in episodic memory from the old tasks. Surprisingly, this simplistic approach only requires to retrain a small amount of data in order to outperform previous methods in knowledge retention. We demonstrate the superiority of our method to the previous ones in two different continual learning settings on popular benchmarks, as well as a new continual learning problem where tasks are designed to be more dissimilar. …

Context-Aware Self-Attention Network
Self-attention model have shown its flexibility in parallel computation and the effectiveness on modeling both long- and short-term dependencies. However, it calculates the dependencies between representations without considering the contextual information, which have proven useful for modeling dependencies among neural representations in various natural language tasks. In this work, we focus on improving self-attention networks through capturing the richness of context. To maintain the simplicity and flexibility of the self-attention networks, we propose to contextualize the transformations of the query and key layers, which are used to calculates the relevance between elements. Specifically, we leverage the internal representations that embed both global and deep contexts, thus avoid relying on external resources. Experimental results on WMT14 English-German and WMT17 Chinese-English translation tasks demonstrate the effectiveness and universality of the proposed methods. Furthermore, we conducted extensive analyses to quantity how the context vectors participate in the self-attention model. …

Structured Generative Adversarial Network (SGAN)
We study the problem of conditional generative modeling based on designated semantics or structures. Existing models that build conditional generators either require massive labeled instances as supervision or are unable to accurately control the semantics of generated samples. We propose structured generative adversarial networks (SGANs) for semi-supervised conditional generative modeling. SGAN assumes the data x is generated conditioned on two independent latent variables: y that encodes the designated semantics, and z that contains other factors of variation. To ensure disentangled semantics in y and z, SGAN builds two collaborative games in the hidden space to minimize the reconstruction error of y and z, respectively. Training SGAN also involves solving two adversarial games that have their equilibrium concentrating at the true joint data distributions p(x, z) and p(x, y), avoiding distributing the probability mass diffusely over data space that MLE-based methods may suffer. We assess SGAN by evaluating its trained networks, and its performance on downstream tasks. We show that SGAN delivers a highly controllable generator, and disentangled representations; it also establishes start-of-the-art results across multiple datasets when applied for semi-supervised image classification (1.27%, 5.73%, 17.26% error rates on MNIST, SVHN and CIFAR-10 using 50, 1000 and 4000 labels, respectively). Benefiting from the separate modeling of y and z, SGAN can generate images with high visual quality and strictly following the designated semantic, and can be extended to a wide spectrum of applications, such as style transfer. …

Parallel Temporal Neural Coding Network
Temporal models based on recurrent neural networks have proven to be quite powerful in a wide variety of applications, including language modeling and speech processing. However, to train these models, one relies on back-propagation through time, which entails unfolding the network over many time steps, making the process of conducting credit assignment considerably more challenging. Furthermore, the nature of back-propagation itself does not permit the use of non-differentiable activation functions and is inherently sequential, making parallelization of the underlying training process very difficult. In this work, we propose the Parallel Temporal Neural Coding Network, a biologically inspired model trained by the local learning algorithm known as Local Representation Alignment, that aims to resolve the difficulties and problems that plague recurrent networks trained by back-propagation through time. Most notably, this architecture requires neither unrolling nor the derivatives of its internal activation functions. We compare our model and learning procedure to other online back-propagation-through-time alternatives (which also tend to be computationally expensive), including real-time recurrent learning, echo state networks, and unbiased online recurrent optimization, and show that it outperforms them on sequence modeling benchmarks such as Bouncing MNIST, a new benchmark we call Bouncing NotMNIST, and Penn Treebank. Notably, our approach can, in some instances, even outperform full back-propagation through time itself as well as variants such as sparse attentive back-tracking. Furthermore, we present promising experimental results that demonstrate our model’s ability to conduct zero-shot adaptation. …

BAyesian Least Squares Optimization with Nonnegative L1-norm constraint (BALSON)
A Bayesian approach termed BAyesian Least Squares Optimization with Nonnegative L1-norm constraint (BALSON) is proposed. The error distribution of data fitting is described by Gaussian likelihood. The parameter distribution is assumed to be a Dirichlet distribution. With the Bayes rule, searching for the optimal parameters is equivalent to finding the mode of the posterior distribution. In order to explicitly characterize the nonnegative L1-norm constraint of the parameters, we further approximate the true posterior distribution by a Dirichlet distribution. We estimate the statistics of the approximating Dirichlet posterior distribution by sampling methods. Four sampling methods have been introduced. With the estimated posterior distributions, the original parameters can be effectively reconstructed in polynomial fitting problems, and the BALSON framework is found to perform better than conventional methods. …

Summit
Deep learning is increasingly used in decision-making tasks. However, understanding how neural networks produce final predictions remains a fundamental challenge. Existing work on interpreting neural network predictions for images often focuses on explaining predictions for single images or neurons. As predictions are often computed based off of millions of weights that are optimized over millions of images, such explanations can easily miss a bigger picture. We present Summit, the first interactive system that scalably and systematically summarizes and visualizes what features a deep learning model has learned and how those features interact to make predictions. Summit introduces two new scalable summarization techniques: (1) activation aggregation discovers important neurons, and (2) neuron-influence aggregation identifies relationships among such neurons. Summit combines these techniques to create the novel attribution graph that reveals and summarizes crucial neuron associations and substructures that contribute to a model’s outcomes. Summit scales to large data, such as the ImageNet dataset with 1.2M images, and leverages neural network feature visualization and dataset examples to help users distill large, complex neural network models into compact, interactive visualizations. We present neural network exploration scenarios where Summit helps us discover multiple surprising insights into a state-of-the-art image classifier’s learned representations and informs future neural network architecture design. The Summit visualization runs in modern web browsers and is open-sourced. …

Thermodynamic Analytics ToolkIt (TATi)
We describe a TensorFlow-based library for posterior sampling and exploration in machine learning applications. TATi, the Thermodynamic Analytics ToolkIt, implements algorithms for 2nd order (underdamped) Langevin dynamics and Hamiltonian Monte Carlo (HMC). It also allows for rapid prototyping of new sampling methods in pure Python and supports an ensemble framework for generating multiple trajectories in parallel, a capability that is demonstrated by the implementation of a recently proposed ensemble preconditioning sampling procedure. In addition to explaining the architecture of TATi and its connections with the TensorFlow framework, this article contains preliminary numerical experiments to explore the efficiency of posterior sampling strategies in ML applications, in comparison to standard training strategies. We provide a glimpse of the potential of the new toolkit by studying (and visualizing) the loss landscape of a neural network applied to the MNIST hand-written digits data set. …

Deep Inverse Optimization
Given a set of observations generated by an optimization process, the goal of inverse optimization is to determine likely parameters of that process. We cast inverse optimization as a form of deep learning. Our method, called deep inverse optimization, is to unroll an iterative optimization process and then use backpropagation to learn parameters that generate the observations. We demonstrate that by backpropagating through the interior point algorithm we can learn the coefficients determining the cost vector and the constraints, independently or jointly, for both non-parametric and parametric linear programs, starting from one or multiple observations. With this approach, inverse optimization can leverage concepts and algorithms from deep learning. …

Hierarchically Supervised Latent Dirichlet Allocation (HSLDA)
We introduce hierarchically supervised latent Dirichlet allocation (HSLDA), a model for hierarchically and multiply labeled bag-of-word data. Examples of such data include web pages and their placement in directories, product descriptions and associated categories from product hierarchies, and free-text clinical records and their assigned diagnosis codes. Out-of-sample label prediction is the primary goal of this work, but improved lower-dimensional representations of the bag-of-word data are also of interest. We demonstrate HSLDA on large-scale data from clinical document labeling and retail product categorization tasks. We show that leveraging the structure from hierarchical labels improves out-of-sample label prediction substantially when compared to models that do not. …

Structural Agnostic Model (SAM)
We present the Structural Agnostic Model (SAM), a framework to estimate end-to-end non-acyclic causal graphs from observational data. In a nutshell, SAM implements an adversarial game in which a separate model generates each variable, given real values from all others. In tandem, a discriminator attempts to distinguish between the joint distributions of real and generated samples. Finally, a sparsity penalty forces each generator to consider only a small subset of the variables, yielding a sparse causal graph. SAM scales easily to hundreds variables. Our experiments show the state-of-the-art performance of SAM on discovering causal structures and modeling interventions, in both acyclic and non-acyclic graphs. …

Hierarchical Attention Mechanism (Ham)
Attention mechanisms in sequence to sequence models have shown great ability and wonderful performance in various natural language processing (NLP) tasks, such as sentence embedding, text generation, machine translation, machine reading comprehension, etc. Unfortunately, existing attention mechanisms only learn either high-level or low-level features. In this paper, we think that the lack of hierarchical mechanisms is a bottleneck in improving the performance of the attention mechanisms, and propose a novel Hierarchical Attention Mechanism (Ham) based on the weighted sum of different layers of a multi-level attention. Ham achieves a state-of-the-art BLEU score of 0.26 on Chinese poem generation task and a nearly 6.5% averaged improvement compared with the existing machine reading comprehension models such as BIDAF and Match-LSTM. Furthermore, our experiments and theorems reveal that Ham has greater generalization and representation ability than existing attention mechanisms. …

Quantized Compressive K-Means
The recent framework of compressive statistical learning aims at designing tractable learning algorithms that use only a heavily compressed representation-or sketch-of massive datasets. Compressive K-Means (CKM) is such a method: it estimates the centroids of data clusters from pooled, non-linear, random signatures of the learning examples. While this approach significantly reduces computational time on very large datasets, its digital implementation wastes acquisition resources because the learning examples are compressed only after the sensing stage. The present work generalizes the sketching procedure initially defined in Compressive K-Means to a large class of periodic nonlinearities including hardware-friendly implementations that compressively acquire entire datasets. This idea is exemplified in a Quantized Compressive K-Means procedure, a variant of CKM that leverages 1-bit universal quantization (i.e. retaining the least significant bit of a standard uniform quantizer) as the periodic sketch nonlinearity. Trading for this resource-efficient signature (standard in most acquisition schemes) has almost no impact on the clustering performances, as illustrated by numerical experiments. …

Spectral Clustering using Deep Neural Networks (SpectralNet)
Spectral clustering is a leading and popular technique in unsupervised data analysis. Two of its major limitations are scalability and generalization of the spectral embedding (i.e., out-of-sample-extension). In this paper we introduce a deep learning approach to spectral clustering that overcomes the above shortcomings. Our network, which we call SpectralNet, learns a map that embeds input data points into the eigenspace of their associated graph Laplacian matrix and subsequently clusters them. We train SpectralNet using a procedure that involves constrained stochastic optimization. Stochastic optimization allows it to scale to large datasets, while the constraints, which are implemented using a special-purpose output layer, allow us to keep the network output orthogonal. Moreover, the map learned by SpectralNet naturally generalizes the spectral embedding to unseen data points. To further improve the quality of the clustering, we replace the standard pairwise Gaussian affinities with affinities leaned from unlabeled data using a Siamese network. Additional improvement can be achieved by applying the network to code representations produced, e.g., by standard autoencoders. Our end-to-end learning procedure is fully unsupervised. In addition, we apply VC dimension theory to derive a lower bound on the size of SpectralNet. State-of-the-art clustering results are reported on the Reuters dataset. Our implementation is publicly available at https://…/SpectralNet . …