Attentive Dynamics Model (ADM)
This paper investigates whether learning contingency-awareness and controllable aspects of an environment can lead to better exploration in reinforcement learning. To investigate this question, we consider an instantiation of this hypothesis evaluated on the Arcade Learning Element (ALE). In this study, we develop an attentive dynamics model (ADM) that discovers controllable elements of the observations, which are often associated with the location of the character in Atari games. The ADM is trained in a self-supervised fashion to predict the actions taken by the agent. The learned contingency information is used as a part of the state representation for exploration purposes. We demonstrate that combining A2C with count-based exploration using our representation achieves impressive results on a set of notoriously challenging Atari games due to sparse rewards. For example, we report a state-of-the-art score of >6600 points on Montezuma’s Revenge without using expert demonstrations, explicit high-level information (e.g., RAM states), or supervised data. Our experiments confirm that indeed contingency-awareness is an extremely powerful concept for tackling exploration problems in reinforcement learning and opens up interesting research questions for further investigations. …

Zest
Programs expecting structured inputs often consist of both a syntactic analysis stage in which raw input is parsed into an internal data structure and a semantic analysis stage which conducts checks on this data structure and executes the core logic of the program. Existing random testing methodologies, like coverage-guided fuzzing (CGF) and generator-based fuzzing, tend to produce inputs that are rejected early in one of these two stages. We propose Zest, a random testing methodology that effectively explores the semantic analysis stages of such programs. Zest combines two key innovations to achieve this. First, we introduce validity fuzzing, which biases CGF towards generating semantically valid inputs. Second, we introduce parametric generators, which convert input from a simple parameter domain, such as a sequence of numbers, into a more structured domain, such as syntactically valid XML. These generators enable parameter-level mutations to map to structural mutations in syntactically valid test inputs. We implement Zest in Java and evaluate it against AFL and QuickCheck, popular CGF and generator-based fuzzing tools, on six real-world benchmarks: Apache Maven, Ant, and BCEL, ScalaChess, the Google Closure compiler, and Mozilla Rhino. We find that Zest achieves the highest coverage of the semantic analysis stage for five of these benchmarks. Further, we find 18 new bugs across the benchmarks, including 7 bugs that are uniquely found by Zest. …

Fisher-Bures Adversary Graph Convolutional Network
In a graph convolutional network, we assume that the graph $G$ is generated with respect to some observation noise. We make small random perturbations $\Delta{}G$ of the graph and try to improve generalization. Based on quantum information geometry, we can have quantitative measurements on the scale of $\Delta{}G$. We try to maximize the intrinsic scale of the permutation with a small budget while minimizing the loss based on the perturbed $G+\Delta{G}$. Our proposed model can consistently improve graph convolutional networks on semi-supervised node classification tasks with reasonable computational overhead. We present two different types of geometry on the manifold of graphs: one is for measuring the intrinsic change of a graph; the other is for measuring how such changes can affect externally a graph neural network. These new analytical tools will be useful in developing a good understanding of graph neural networks and fostering new techniques. …

Majority-CRF
We explore active learning (AL) utterance selection for improving the accuracy of new underrepresented domains in a natural language understanding (NLU) system. Moreover, we propose an AL algorithm called Majority-CRF that uses an ensemble of classification and sequence labeling models to guide utterance selection for annotation. Experiments with three domains show that Majority-CRF achieves 6.6%-9% relative error rate reduction compared to random sampling with the same annotation budget, and statistically significant improvements compared to other AL approaches. Additionally, case studies with human-in-the-loop AL on six new domains show 4.6%-9% improvement on an existing NLU system. …

Statistical Learning Theory (SLT)
Statistical learning theory is a framework for machine learning drawing from the fields of statistics and functional analysis. Statistical learning theory deals with the problem of finding a predictive function based on data. Statistical learning theory has led to successful applications in fields such as computer vision, speech recognition, bioinformatics and baseball. …

Social-Relation based Centrality (SoReC)
Mobile Social Networks (MSNs) have been evolving and enabling various fields in recent years. Recent advances in mobile edge computing, caching, and device-to-device communications, can have significant impacts on 5G systems. In those settings, identifying central users is crucial. It can provide important insights into designing and deploying diverse services and applications. However, it is challenging to evaluate the centrality of nodes in MSNs with dynamic environments. In this paper, we propose a Social-Relation based Centrality (SoReC) measure, in which social network information is used to quantify the influence of each user in MSNs. We first introduce a new metric to estimate direct social relations among users via direct contacts, and then extend the metric to explore indirect social relations among users bridging by the third parties. Based on direct and indirect social relations, we detect the influence spheres of users and quantify their influence in the networks. Simulations on real-world networks show that the proposed measure can perform well in identifying future influential users in MSNs. …

Conditional Random Fields as Recurrent Neural Networks (CRF-RNN)
Pixel-level labelling tasks, such as semantic segmentation, play a central role in image understanding. Recent approaches have attempted to harness the capabilities of deep learning techniques for image recognition to tackle pixel-level labelling tasks. One central issue in this methodology is the limited capacity of deep learning techniques to delineate visual objects. To solve this problem, we introduce a new form of convolutional neural network that combines the strengths of Convolutional Neural Networks (CNNs) and Conditional Random Fields (CRFs)-based probabilistic graphical modelling. To this end, we formulate Conditional Random Fields as Recurrent Neural Networks. This network, called CRF-RNN, is then plugged in as a part of a CNN to obtain a deep network that has desirable properties of both CNNs and CRFs. Importantly, our system fully integrates CRF modelling with CNNs, making it possible to train the whole deep network end-to-end with the usual back-propagation algorithm, avoiding offline post-processing methods for object delineation.
GitXiv …

Levy-Attack
Developing techniques for adversarial attack and defense is an important research field for establishing reliable machine learning and its applications. Many existing methods employ Gaussian random variables for exploring the data space to find the most adversarial (for attacking) or least adversarial (for defense) point. However, the Gaussian distribution is not necessarily the optimal choice when the exploration is required to follow the complicated structure that most real-world data distributions exhibit. In this paper, we investigate how statistics of random variables affect such random walk exploration. Specifically, we generalize the Boundary Attack, a state-of-the-art black-box decision based attacking strategy, and propose the L\’evy-Attack, where the random walk is driven by symmetric $\alpha$-stable random variables. Our experiments on MNIST and CIFAR10 datasets show that the L\’evy-Attack explores the image data space more efficiently, and significantly improves the performance. Our results also give an insight into the recently found fact in the whitebox attacking scenario that the choice of the norm for measuring the amplitude of the adversarial patterns is essential. …

Multi-model and Multi-level Knowledge Distillation (M2KD)
Incremental learning targets at achieving good performance on new categories without forgetting old ones. Knowledge distillation has been shown critical in preserving the performance on old classes. Conventional methods, however, sequentially distill knowledge only from the last model, leading to performance degradation on the old classes in later incremental learning steps. In this paper, we propose a multi-model and multi-level knowledge distillation strategy. Instead of sequentially distilling knowledge only from the last model, we directly leverage all previous model snapshots. In addition, we incorporate an auxiliary distillation to further preserve knowledge encoded at the intermediate feature levels. To make the model more memory efficient, we adapt mask based pruning to reconstruct all previous models with a small memory footprint. Experiments on standard incremental learning benchmarks show that our method preserves the knowledge on old classes better and improves the overall performance over standard distillation techniques. …

A Swift Approximate Pattern-Miner
While there has been a tremendous interest in processing data that has an underlying graph structure, existing distributed graph processing systems take several minutes or even hours to mine simple patterns on graphs. This paper presents ASAP, a fast, approximate computation engine for graph pattern mining. ASAP leverages state-of-the-art results in graph approximation theory, and extends it to general graph patterns in distributed settings. To enable the users to navigate the trade-off between the result accuracy and latency, we propose a novel approach to build the Error-Latency Profile (ELP) for a given computation. We have implemented ASAP on a general-purpose distributed dataflow platform, and evaluated it extensively on several graph patterns. Our experimental results show that ASAP outperforms existing exact pattern mining solutions by up to 77×. Further, ASAP can scale to graphs with billions of edges without the need for large clusters. …

Res-SE-Net
One of the ways to train deep neural networks effectively is to use residual connections. Residual connections can be classified as being either identity connections or bridge-connections with a reshaping convolution. Empirical observations on CIFAR-10 and CIFAR-100 datasets using a baseline Resnet model, with bridge-connections removed, have shown a significant reduction in accuracy. This reduction is due to lack of contribution, in the form of feature maps, by the bridge-connections. Hence bridge-connections are vital for Resnet. However, all feature maps in the bridge-connections are considered to be equally important. In this work, an upgraded architecture ‘Res-SE-Net’ is proposed to further strengthen the contribution from the bridge-connections by quantifying the importance of each feature map and weighting them accordingly using Squeeze-and-Excitation (SE) block. It is demonstrated that Res-SE-Net generalizes much better than Resnet and SE-Resnet on the benchmark CIFAR-10 and CIFAR-100 datasets. …

Learning Using Statistical Invariants (LUSI)
This paper introduces a new learning paradigm, called Learning Using Statistical Invariants (LUSI), which is different from the classical one. In a classical paradigm, the learning machine constructs a classification rule that minimizes the probability of expected error; it is data-driven model of learning. In the LUSI paradigm, in order to construct the desired classification function, a learning machine computes statistical invariants that are specific for the problem, and then minimizes the expected error in a way that preserves these invariants; it is thus both data- and invariant-driven learning. From a mathematical point of view, methods of the classical paradigm employ mechanisms of strong convergence of approximations to the desired function, whereas methods of the new paradigm employ both strong and weak convergence mechanisms. This can significantly increase the rate of convergence. …

EventKG
One of the key requirements to facilitate the semantic analytics of information regarding contemporary and historical events on the Web, in the news and in social media is the availability of reference knowledge repositories containing comprehensive representations of events, entities and temporal relations. Existing knowledge graphs, with popular examples including DBpedia, YAGO and Wikidata, focus mostly on entity-centric information and are insufficient in terms of their coverage and completeness with respect to events and temporal relations. In this article we address this limitation, formalise the concept of a temporal knowledge graph and present its instantiation – EventKG. EventKG is a multilingual event-centric temporal knowledge graph that incorporates over 690 thousand events and over 2.3 million temporal relations obtained from several large-scale knowledge graphs and semi-structured sources and makes them available through a canonical RDF representation. Whereas popular entities often possess hundreds of relations within a temporal knowledge graph such as EventKG, generating a concise overview of the most important temporal relations for a given entity is a challenging task. In this article we demonstrate an application of EventKG to biographical timeline generation, where we adopt a distant supervision method to identify relations most relevant for an entity biography. Our evaluation results provide insights on the characteristics of EventKG and demonstrate the effectiveness of the proposed biographical timeline generation method. …

Kernel Normalized Least-Mean Square (KNLMS)
In the last decade, a considerable research effort has been devoted to developing adaptive algorithms based on kernel functions. One of the main features of these algorithms is that they form a family of universal approximation techniques, solving problems with nonlinearities elegantly. In this paper, we present data-selective adaptive kernel normalized least-mean square (KNLMS) algorithms that can increase their learning rate and reduce their computational complexity. In fact, these methods deal with kernel expansions, creating a growing structure also known as the dictionary, whose size depends on the number of observations and their innovation. The algorithms described herein use an adaptive step-size to accelerate the learning and can offer an excellent tradeoff between convergence speed and steady state, which allows them to solve nonlinear filtering and estimation problems with a large number of parameters without requiring a large computational cost. The data-selective update scheme also limits the number of operations performed and the size of the dictionary created by the kernel expansion, saving computational resources and dealing with one of the major problems of kernel adaptive algorithms. A statistical analysis is carried out along with a computational complexity analysis of the proposed algorithms. Simulations show that the proposed KNLMS algorithms outperform existing algorithms in examples of nonlinear system identification and prediction of a time series originating from a nonlinear difference equation. …

OPENMENDEL
Statistical methods for genomewide association studies (GWAS) continue to improve. However, the increasing volume and variety of genetic and genomic data make computational speed and ease of data manipulation mandatory in future software. In our view, a collaborative effort of statistical geneticists is required to develop open source software targeted to genetic epidemiology. Our attempt to meet this need is called the OPENMENDELproject (https://openmendel.github.io ). It aims to (1) enable interactive and reproducible analyses with informative intermediate results, (2) scale to big data analytics, (3) embrace parallel and distributed computing, (4) adapt to rapid hardware evolution, (5) allow cloud computing, (6) allow integration of varied genetic data types, and (7) foster easy communication between clinicians, geneticists, statisticians, and computer scientists. This article reviews and makes recommendations to the genetic epidemiology community in the context of the OPENMENDEL project. …

Cover Complexity (CC)
The accuracy of deep learning, i.e., deep neural networks, can be characterized by dividing the total error into three main types: approximation error, optimization error, and generalization error. Whereas there are some satisfactory answers to the problems of approximation and optimization, much less is known about the theory of generalization. Most existing theoretical works for generalization fail to explain the performance of neural networks in practice. To derive a meaningful bound, we study the generalization error of neural networks for classification problems in terms of data distribution and neural network smoothness. We introduce the cover complexity (CC) to measure the difficulty of learning a data set and the inverse of modules of continuity to quantify neural network smoothness. A quantitative bound for expected accuracy/error is derived by considering both the CC and neural network smoothness. We validate our theoretical results by several data sets of images. The numerical results verify that the expected error of trained networks scaled with the square root of the number of classes has a linear relationship with respect to the CC. In addition, we observe a clear consistency between test loss and neural network smoothness during the training process. …

Neural Network Quine
Self-replication is a key aspect of biological life that has been largely overlooked in Artificial Intelligence systems. Here we describe how to build and train self-replicating neural networks. The network replicates itself by learning to output its own weights. The network is designed using a loss function that can be optimized with either gradient-based or non-gradient-based methods. We also describe a method we call regeneration to train the network without explicit optimization, by injecting the network with predictions of its own parameters. The best solution for a self-replicating network was found by alternating between regeneration and optimization steps. Finally, we describe a design for a self-replicating neural network that can solve an auxiliary task such as MNIST image classification. We observe that there is a trade-off between the network’s ability to classify images and its ability to replicate, but training is biased towards increasing its specialization at image classification at the expense of replication. This is analogous to the trade-off between reproduction and other tasks observed in nature. We suggest that a self-replication mechanism for artificial intelligence is useful because it introduces the possibility of continual improvement through natural selection. …

Time-Domain Audio Separation Network (TasNet)
Robust speech processing in multitalker acoustic environments requires automatic speech separation. While single-channel, speaker-independent speech separation methods have recently seen great progress, the accuracy, latency, and computational cost of speech separation remain insufficient. The majority of the previous methods have formulated the separation problem through the time-frequency representation of the mixed signal, which has several drawbacks, including the decoupling of the phase and magnitude of the signal, the suboptimality of spectrogram representations for speech separation, and the long latency in calculating the spectrogram. To address these shortcomings, we propose the time-domain audio separation network (TasNet), which is a deep learning autoencoder framework for time-domain speech separation. TasNet uses a convolutional encoder to create a representation of the signal that is optimized for extracting individual speakers. Speaker extraction is achieved by applying a weighting function (mask) to the encoder output. The modified encoder representation is then inverted to the sound waveform using a linear decoder. The masks are found using a temporal convolutional network consisting of dilated convolutions, which allow the network to model the long-term dependencies of the speech signal. This end-to-end speech separation algorithm significantly outperforms previous time-frequency methods in terms of separating speakers in mixed audio, even when compared to the separation accuracy achieved with the ideal time-frequency mask of the speakers. In addition, TasNet has a smaller model size and a shorter minimum latency, making it a suitable solution for both offline and real-time speech separation applications. This study therefore represents a major step toward actualizing speech separation for real-world speech processing technologies. …

greta
greta lets us build statistical models interactively in R, and then sample from them by MCMC. We build greta models with greta array objects, which behave much like R’s array, matrix and vector objects for numeric data. Like those numeric data objects, greta arrays can be manipulated with functions and mathematical operators to create new greta arrays. The key difference between greta arrays and numeric data objects is that when you do something to a greta array, greta doesn’t calculate the values of the new greta array. Instead, it just remembers what operation to do, and works out the size and shape of the result. …

ELM with Local Connections (ELM-LC)
This paper is concerned with the sparsification of the input-hidden weights of ELM (Extreme Learning Machine). For ordinary feedforward neural networks, the sparsification is usually done by introducing certain regularization technique into the learning process of the network. But this strategy can not be applied for ELM, since the input-hidden weights of ELM are supposed to be randomly chosen rather than to be learned. To this end, we propose a modified ELM, called ELM-LC (ELM with local connections), which is designed for the sparsification of the input-hidden weights as follows: The hidden nodes and the input nodes are divided respectively into several corresponding groups, and an input node group is fully connected with its corresponding hidden node group, but is not connected with any other hidden node group. As in the usual ELM, the hidden-input weights are randomly given, and the hidden-output weights are obtained through a least square learning. In the numerical simulations on some benchmark problems, the new ELM-CL behaves better than the traditional ELM. …

Independent Component Analysis (ICA)
In signal processing, independent component analysis (ICA) is a computational method for separating a multivariate signal into additive subcomponents. This is done by assuming that the subcomponents are non-Gaussian signals and that they are statistically independent from each other. ICA is a special case of blind source separation. A common example application is the ‘cocktail party problem’ of listening in on one person’s speech in a noisy room.
➘ “Principal Component Analysis” …

Directed Exploration Learning (DEL)
We address reinforcement learning problems with finite state and action spaces where the underlying MDP has some known structure that could be potentially exploited to minimize the exploration of suboptimal (state, action) pairs. For any arbitrary structure, we derive problem-specific regret lower bounds satisfied by any learning algorithm. These lower bounds are made explicit for unstructured MDPs and for those whose transition probabilities and average reward function are Lipschitz continuous w.r.t. the state and action. For Lipschitz MDPs, the bounds are shown not to scale with the sizes $S$ and $A$ of the state and action spaces, i.e., they are smaller than $c \log T$ where $T$ is the time horizon and the constant $c$ only depends on the Lipschitz structure, the span of the bias function, and the minimal action sub-optimality gap. This contrasts with unstructured MDPs where the regret lower bound typically scales as $SA \log T$ . We devise DEL (Directed Exploration Learning), an algorithm that matches our regret lower bounds. We further simplify the algorithm for Lipschitz MDPs, and show that the simplified version is still able to efficiently exploit the structure. …

Clustering-Correcting Code
A new family of codes, called clustering-correcting codes, is presented in this paper. This family of codes is motivated by the special structure of data that is stored in DNA-based storage systems. The data stored in these systems has the form of unordered sequences, also called strands, and every strand is synthesized thousands to millions of times, where some of these copies are read back during sequencing. Due to the unordered structure of the strands, an important task in the decoding process is to place them in their correct order. This is usually accomplished by allocating a part of the strand for an index. However, in the presence of errors in the index field, important information on the order of the strands may be lost. Clustering-correcting codes ensure that if the distance between the index fields of two strands is small, then there will be a large distance between their data fields. It is shown how this property enables to place the strands together in their correct clusters even in the presence of errors. We present lower and upper bounds on the size of clustering-correcting codes and an explicit construction of these codes which uses only a single bit of redundancy. …

Multi-Target Filtering and Tracking (MTFT)
Defining a multi-target motion model, which is an important step of tracking algorithms, can be very challenging. Using fixed models (as in several generative Bayesian algorithms, such as Kalman filters) can fail to accurately predict sophisticated target motions. On the other hand, sequential learning of the motion model (for example, using recurrent neural networks) can be computationally complex and difficult due to the variable unknown number of targets. In this paper, we propose a multi-target filtering and tracking (MTFT) algorithm which learns the motion model, simultaneously for all targets, from an implicitly represented state map and performs spatio-temporal data prediction. To this end, the multi-target state is modelled over a continuous hypothetical target space, using random finite sets and Gaussian mixture probability hypothesis density formulations. The prediction step is recursively performed using a deep convolutional recurrent neural network with a long short-term memory architecture, which is trained as a regression block, on the fly, over ‘probability density difference’ maps. Our approach is evaluated over widely used pedestrian tracking benchmarks, remarkably outperforming state-of-the-art multi-target filtering algorithms, while giving competitive results when compared with other tracking approaches. …

Cross Local Intrinsic Dimensionality (LID)
Generative Adversarial Networks (GANs) are an elegant mechanism for data generation. However, a key challenge when using GANs is how to best measure their ability to generate realistic data. In this paper, we demonstrate that an intrinsic dimensional characterization of the data space learned by a GAN model leads to an effective evaluation metric for GAN quality. In particular, we propose a new evaluation measure, CrossLID, that assesses the local intrinsic dimensionality (LID) of real-world data with respect to neighborhoods found in GAN-generated samples. Intuitively, CrossLID measures the degree to which manifolds of two data distributions coincide with each other. In experiments on 4 benchmark image datasets, we compare our proposed measure to several state-of-the-art evaluation metrics. Our experiments show that CrossLID is strongly correlated with the progress of GAN training, is sensitive to mode collapse, is robust to small-scale noise and image transformations, and robust to sample size. Furthermore, we show how CrossLID can be used within the GAN training process to improve generation quality. …

Context-Aware Detection Network (CAD-Net)
Accurate and robust detection of multi-class objects in optical remote sensing images is essential to many real-world applications such as urban planning, traffic control, searching and rescuing, etc. However, state-of-the-art object detection techniques designed for images captured using ground-level sensors usually experience a sharp performance drop when directly applied to remote sensing images, largely due to the object appearance differences in remote sensing images in term of sparse texture, low contrast, arbitrary orientations, large scale variations, etc. This paper presents a novel object detection network (CAD-Net) that exploits attention-modulated features as well as global and local contexts to address the new challenges in detecting objects from remote sensing images. The proposed CAD-Net learns global and local contexts of objects by capturing their correlations with the global scene (at scene-level) and the local neighboring objects or features (at object-level), respectively. In addition, it designs a spatial-and-scale-aware attention module that guides the network to focus on more informative regions and features as well as more appropriate feature scales. Experiments over two publicly available object detection datasets for remote sensing images demonstrate that the proposed CAD-Net achieves superior detection performance. The implementation codes will be made publicly available for facilitating future researches. …

Reversible Recurrent Neural Network
Recurrent neural networks (RNNs) provide state-of-the-art performance in processing sequential data but are memory intensive to train, limiting the flexibility of RNN models which can be trained. Reversible RNNs—RNNs for which the hidden-to-hidden transition can be reversed—offer a path to reduce the memory requirements of training, as hidden states need not be stored and instead can be recomputed during backpropagation. We first show that perfectly reversible RNNs, which require no storage of the hidden activations, are fundamentally limited because they cannot forget information from their hidden state. We then provide a scheme for storing a small number of bits in order to allow perfect reversal with forgetting. Our method achieves comparable performance to traditional models while reducing the activation memory cost by a factor of 10–15. We extend our technique to attention-based sequence-to-sequence models, where it maintains performance while reducing activation memory cost by a factor of 5–10 in the encoder, and a factor of 10–15 in the decoder. …

d-Orientable Deletion
A graph is $d$-orientable if its edges can be oriented so that the maximum in-degree of the resulting digraph is at most $d$. $d$-orientability is a well-studied concept with close connections to fundamental graph-theoretic notions and applications as a load balancing problem. In this paper we consider the d-ORIENTABLE DELETION problem: given a graph $G=(V,E)$, delete the minimum number of vertices to make $G$ $d$-orientable. We contribute a number of results that improve the state of the art on this problem. Specifically: – We show that the problem is W[2]-hard and $\log n$-inapproximable with respect to $k$, the number of deleted vertices. This closes the gap in the problem’s approximability. – We completely characterize the parameterized complexity of the problem on chordal graphs: it is FPT parameterized by $d+k$, but W-hard for each of the parameters $d,k$ separately. – We show that, under the SETH, for all $d,\epsilon$, the problem does not admit a $(d+2-\epsilon)^{tw}$, algorithm where $tw$ is the graph’s treewidth, resolving as a special case an open problem on the complexity of PSEUDOFOREST DELETION. – We show that the problem is W-hard parameterized by the input graph’s clique-width. Complementing this, we provide an algorithm running in time $d^{O(d\cdot cw)}$, showing that the problem is FPT by $d+cw$, and improving the previously best known algorithm for this case. …

neuralRank
Widespread applications of deep learning have led to a plethora of pre-trained neural network models for common tasks. Such models are often adapted from other models via transfer learning. The models may have varying training sets, training algorithms, network architectures, and hyper-parameters. For a given application, what isthe most suitable model in a model repository? This is a critical question for practical deployments but it has not received much attention. This paper introduces the novel problem of searching and ranking models based on suitability relative to a target dataset and proposes a ranking algorithm called \textit{neuralRank}. The key idea behind this algorithm is to base model suitability on the discriminating power of a model, using a novel metric to measure it. With experimental results on the MNIST, Fashion, and CIFAR10 datasets, we demonstrate that (1) neuralRank is independent of the domain, the training set, or the network architecture and (2) that the models ranked highly by neuralRank ranking tend to have higher model accuracy in practice. …

NSCaching
Knowledge Graph (KG) embedding is a fundamental problem in data mining research with many real-world applications. It aims to encode the entities and relations in the graph into low dimensional vector space, which can be used for subsequent algorithms. Negative sampling, which samples negative triplets from non-observed ones in the training data, is an important step in KG embedding. Recently, generative adversarial network (GAN), has been introduced in negative sampling. By sampling negative triplets with large scores, these methods avoid the problem of vanishing gradient and thus obtain better performance. However, using GAN makes the original model more complex and hard to train, where reinforcement learning must be used. In this paper, motivated by the observation that negative triplets with large scores are important but rare, we propose to directly keep track of them with the cache. However, how to sample from and update the cache are two important questions. We carefully design the solutions, which are not only efficient but also achieve a good balance between exploration and exploitation. In this way, our method acts as a ‘distilled’ version of previous GA-based methods, which does not waste training time on additional parameters to fit the full distribution of negative triplets. The extensive experiments show that our method can gain significant improvement in various KG embedding models, and outperform the state-of-the-art negative sampling methods based on GAN. …

Amazon SageMaker Ground Truth
In 1959, Arthur Samuel defined machine learning as a ‘field of study that gives computers the ability to learn without being explicitly programmed’. However, there is no deus ex machina: the learning process requires an algorithm (‘how to learn’) and a training dataset (‘what to learn from’). Today, most machine learning tasks use a technique called supervised learning: an algorithm learns patterns or behaviours from a labeled dataset. A labeled dataset containing data samples as well as the correct answer for each one of them, aka ‘ground truth’. Depending on the problem at hand, one could use labeled images (‘this is a dog’, ‘this is a cat’), labeled text (‘this is spam’, ‘this isn’t’), etc. Fortunately, developers and data scientists can now rely on a vast collection of off-the-shelf algorithms (as illustrated by the built-in algorithms in Amazon SageMaker) and of reference datasets. Deep learning has popularized image datasets such as MNIST, CIFAR-10 or ImageNet, and more are also available for tasks like machine translation or text classification. These reference datasets are extremely useful for beginners and experienced practitioners alike, but a lot of companies and organizations still need to train machine learning models on their own dataset: think about medical imaging, autonomous driving, etc. …

Short-Time Fourier Neural Network (STFNet)
Recent advances in deep learning motivate the use of deep neural networks in Internet-of-Things (IoT) applications. These networks are modelled after signal processing in the human brain, thereby leading to significant advantages at perceptual tasks such as vision and speech recognition. IoT applications, however, often measure physical phenomena, where the underlying physics (such as inertia, wireless signal propagation, or the natural frequency of oscillation) are fundamentally a function of signal frequencies, offering better features in the frequency domain. This observation leads to a fundamental question: For IoT applications, can one develop a new brand of neural network structures that synthesize features inspired not only by the biology of human perception but also by the fundamental nature of physics? Hence, in this paper, instead of using conventional building blocks (e.g., convolutional and recurrent layers), we propose a new foundational neural network building block, the Short-Time Fourier Neural Network (STFNet). It integrates a widely-used time-frequency analysis method, the Short-Time Fourier Transform, into data processing to learn features directly in the frequency domain, where the physics of underlying phenomena leave better foot-prints. STFNets bring additional flexibility to time-frequency analysis by offering novel nonlinear learnable operations that are spectral-compatible. Moreover, STFNets show that transforming signals to a domain that is more connected to the underlying physics greatly simplifies the learning process. We demonstrate the effectiveness of STFNets with extensive experiments. STFNets significantly outperform the state-of-the-art deep learning models in all experiments. A STFNet, therefore, demonstrates superior capability as the fundamental building block of deep neural networks for IoT applications for various sensor inputs. …

Joint Maximum Likelihood Estimation (JMLE)
JMLE ‘Joint Maximum Likelihood Estimation’ is also called UCON, ‘Unconditional maximum likelihood estimation’. It was devised by Wright & Panchapakesan, www.rasch.org/memo46.htm. In this formulation, the estimate of the Rasch parameter (for which the observed data are most likely, assuming those data fit the Rasch model) occurs when the observed raw score for the parameter matches the expected raw score. ‘Joint’ means that the estimates for the persons (rows) and items (columns) and rating scale structures (if any) of the data matrix are obtained simultaneously.
➘ “Rasch Model” …

CLEVR-Dialog
Visual Dialog is a multimodal task of answering a sequence of questions grounded in an image, using the conversation history as context. It entails challenges in vision, language, reasoning, and grounding. However, studying these subtasks in isolation on large, real datasets is infeasible as it requires prohibitively-expensive complete annotation of the ‘state’ of all images and dialogs. We develop CLEVR-Dialog, a large diagnostic dataset for studying multi-round reasoning in visual dialog. Specifically, we construct a dialog grammar that is grounded in the scene graphs of the images from the CLEVR dataset. This combination results in a dataset where all aspects of the visual dialog are fully annotated. In total, CLEVR-Dialog contains 5 instances of 10-round dialogs for about 85k CLEVR images, totaling to 4.25M question-answer pairs. We use CLEVR-Dialog to benchmark performance of standard visual dialog models; in particular, on visual coreference resolution (as a function of the coreference distance). This is the first analysis of its kind for visual dialog models that was not possible without this dataset. We hope the findings from CLEVR-Dialog will help inform the development of future models for visual dialog. Our dataset and code will be made public. …

Lazily Aggregated Gradient Coding (LAGC)
Gradient-based distributed learning in Parameter Server (PS) computing architectures is subject to random delays due to straggling worker nodes, as well as to possible communication bottlenecks between PS and workers. Solutions have been recently proposed to separately address these impairments based on the ideas of gradient coding, worker grouping, and adaptive worker selection. This paper provides a unified analysis of these techniques in terms of wall-clock time, communication, and computation complexity measures. Furthermore, in order to combine the benefits of gradient coding and grouping in terms of robustness to stragglers with the communication and computation load gains of adaptive selection, novel strategies, named Lazily Aggregated Gradient Coding (LAGC) and Grouped-LAG (G-LAG), are introduced. Analysis and results show that G-LAG provides the best wall-clock time and communication performance, while maintaining a low computational cost, for two representative distributions of the computing times of the worker nodes. …

Unsupervised Data Augmentation (UDA)
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 $30\%$ 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. …

Bayesian Layer
We describe Bayesian Layers, a module designed for fast experimentation with neural network uncertainty. It extends neural network libraries with layers capturing uncertainty over weights (Bayesian neural nets), pre-activation units (dropout), activations (‘stochastic output layers’), and the function itself (Gaussian processes). With reversible layers, one can also propagate uncertainty from input to output such as for flow-based distributions and constant-memory backpropagation. Bayesian Layers are a drop-in replacement for other layers, maintaining core features that one typically desires for experimentation. As demonstration, we fit a 10-billion parameter ‘Bayesian Transformer’ on 512 TPUv2 cores, which replaces attention layers with their Bayesian counterpart. …

Linear Quadratic Regulator
The theory of optimal control is concerned with operating a dynamic system at minimum cost. The case where the system dynamics are described by a set of linear differential equations and the cost is described by a quadratic function is called the LQ problem. One of the main results in the theory is that the solution is provided by the linear-quadratic regulator (LQR), a feedback controller whose equations are given below. The LQR is an important part of the solution to the LQG (linear-quadratic-Gaussian) problem. Like the LQR problem itself, the LQG problem is one of the most fundamental problems in control theory. …

Unrooted Tree
In a context where trees are supposed to have a root, a tree without any designated root is called a free tree / unrooted tree. …

Speculative Query Planning (Spec-QP)
Organisations store huge amounts of data from multiple heterogeneous sources in the form of Knowledge Graphs (KGs). One of the ways to query these KGs is to use SPARQL queries over a database engine. Since SPARQL follows exact match semantics, the queries may return too few or no results. Recent works have proposed query relaxation where the query engine judiciously replaces a query predicate with similar predicates using weighted relaxation rules mined from the KG. The space of possible relaxations is potentially too large to fully explore and users are typically interested in only top-k results, so such query engines use top-k algorithms for query processing. However, they may still process all the relaxations, many of whose answers do not contribute towards top-k answers. This leads to computation overheads and delayed response times. We propose Spec-QP, a query planning framework that speculatively determines which relaxations will have their results in the top-k answers. Only these relaxations are processed using the top-k operators. We, therefore, reduce the computation overheads and achieve faster response times without adversely affecting the quality of results. We tested Spec-QP over two datasets – XKG and Twitter, to demonstrate the efficiency of our planning framework at reducing runtimes with reasonable accuracy for query engines supporting relaxations. …