Stop Words
In computing, stop words are words which are filtered out before or after processing of natural language data (text). There is not one definite list of stop words which all tools use and such a filter is not always used. Some tools specifically avoid removing them to support phrase search. Any group of words can be chosen as the stop words for a given purpose. For some search engines, these are some of the most common, short function words, such as the, is, at, which, and on. In this case, stop words can cause problems when searching for phrases that include them, particularly in names such as ‘The Who’, ‘The The’, or ‘Take That’. Other search engines remove some of the most common words-including lexical words, such as “want”-from a query in order to improve performance. …

Approximate Exploration
Although exploration in reinforcement learning is well understood from a theoretical point of view, provably correct methods remain impractical. In this paper we study the interplay between exploration and approximation, what we call \emph{approximate exploration}. We first provide results when the approximation is explicit, quantifying the performance of an exploration algorithm, MBIE-EB \citep{strehl2008analysis}, when combined with state aggregation. In particular, we show that this allows the agent to trade off between learning speed and quality of the policy learned. We then turn to a successful exploration scheme in practical, pseudo-count based exploration bonuses \citep{bellemare2016unifying}. We show that choosing a density model implicitly defines an abstraction and that the pseudo-count bonus incentivizes the agent to explore using this abstraction. We find, however, that implicit exploration may result in a mismatch between the approximated value function and exploration bonus, leading to either under- or over-exploration. …

GPU Open Analytics Initiative (GOAI)
Recently, Continuum Analytics, H2O.ai, and MapD announced the formation of the GPU Open Analytics Initiative (GOAI). GOAI-also joined by BlazingDB, Graphistry and the Gunrock project from the University of California, Davis-aims to create open frameworks that allow developers and data scientists to build applications using standard data formats and APIs on GPUs. Bringing standard analytics data formats to GPUs will allow data analytics to be even more efficient, and to take advantage of the high throughput of GPUs. NVIDIA believes this initiative is a key contributor to the continued growth of GPU computing in accelerated analytics. …

StochAstic Recursive grAdient algoritHm (SARAH)
In this paper, we propose a StochAstic Recursive grAdient algoritHm (SARAH), as well as its practical variant SARAH+, as a novel approach to the finite-sum minimization problems. Different from the vanilla SGD and other modern stochastic methods such as SVRG, S2GD, SAG and SAGA, SARAH admits a simple recursive framework for updating stochastic gradient estimates; when comparing to SAG/SAGA, SARAH does not require a storage of past gradients. The linear convergence rate of SARAH is proven under strong convexity assumption. We also prove a linear convergence rate (in the strongly convex case) for an inner loop of SARAH, the property that SVRG does not possess. Numerical experiments demonstrate the efficiency of our algorithm. …

HIRO
Hierarchical reinforcement learning (HRL) is a promising approach to extend traditional reinforcement learning (RL) methods to solve more complex tasks. Yet, the majority of current HRL methods require careful task-specific design and on-policy training, making them difficult to apply in real-world scenarios. In this paper, we study how we can develop HRL algorithms that are general, in that they do not make onerous additional assumptions beyond standard RL algorithms, and efficient, in the sense that they can be used with modest numbers of interaction samples, making them suitable for real-world problems such as robotic control. For generality, we develop a scheme where lower-level controllers are supervised with goals that are learned and proposed automatically by the higher-level controllers. To address efficiency, we propose to use off-policy experience for both higher and lower-level training. This poses a considerable challenge, since changes to the lower-level behaviors change the action space for the higher-level policy, and we introduce an off-policy correction to remedy this challenge. This allows us to take advantage of recent advances in off-policy model-free RL to learn both higher- and lower-level policies using substantially fewer environment interactions than on-policy algorithms. We term the resulting HRL agent HIRO and find that it is generally applicable and highly sample-efficient. Our experiments show that HIRO can be used to learn highly complex behaviors for simulated robots, such as pushing objects and utilizing them to reach target locations, learning from only a few million samples, equivalent to a few days of real-time interaction. In comparisons with a number of prior HRL methods, we find that our approach substantially outperforms previous state-of-the-art techniques. …

Augmented Neural ODE
We show that Neural Ordinary Differential Equations (ODEs) learn representations that preserve the topology of the input space and prove that this implies the existence of functions Neural ODEs cannot represent. To address these limitations, we introduce Augmented Neural ODEs which, in addition to being more expressive models, are empirically more stable, generalize better and have a lower computational cost than Neural ODEs. …

Balancing GAN (BAGAN)
Image classification datasets are often imbalanced, characteristic that negatively affects the accuracy of deeplearning classifiers. In this work we propose balancing GANs (BAGANs) as an augmentation tool to restore balance in imbalanced datasets. This is challenging because the few minority-class images may not be enough to train a GAN. We overcome this issue by including during training all available images of majority and minority classes. The generative model learns useful features from majority classes and uses these to generate images for minority classes. We apply class-conditioning in the latent space to drive the generation process towards a target class. Additionally, we couple GANs with autoencoding techniques to reduce the risk of collapsing toward the generation of few foolish examples. We compare the proposed methodology with state-of-the-art GANs and demonstrate that BAGAN generates images of superior quality when trained with an imbalanced dataset. …

Speaker Diarisation
Speaker diarisation (or diarization) is the process of partitioning an input audio stream into homogeneous segments according to the speaker identity. It can enhance the readability of an automatic speech transcription by structuring the audio stream into speaker turns and, when used together with speaker recognition systems, by providing the speaker’s true identity. It is used to answer the question ‘who spoke when?’ Speaker diarisation is a combination of speaker segmentation and speaker clustering. The first aims at finding speaker change points in an audio stream. The second aims at grouping together speech segments on the basis of speaker characteristics. …

Quadratic Unconstrained Binary Optimization (QUBO)
Gaussian Processes are used in many applications to model spatial phenomena. Within this context, a key issue is to decide the set of locations where to take measurements so as to obtain a better approximation of the underlying function. Current state of the art techniques select such set to minimize the posterior variance of the Gaussian process. We explore the feasibility of solving this problem by proposing a novel Quadratic Unconstrained Binary Optimization (QUBO) model. In recent years this QUBO formulation has gained increasing attention since it represents the input for the specialized quantum annealer D-Wave machines. Hence, our contribution takes an important first step towards the sampling optimization of Gaussian processes in the context of quantum computation. Results of our empirical evaluation shows that the optimum of the QUBO objective function we derived represents a good solution for the above mentioned problem. In fact we are able to obtain comparable and in some cases better results than the widely used submodular technique. …

Point Attention Transformer (PAT)
Geometric deep learning is increasingly important thanks to the popularity of 3D sensors. Inspired by the recent advances in NLP domain, the self-attention transformer is introduced to consume the point clouds. We develop Point Attention Transformers (PATs), using a parameter-efficient Group Shuffle Attention (GSA) to replace the costly Multi-Head Attention. We demonstrate its ability to process size-varying inputs, and prove its permutation equivariance. Besides, prior work uses heuristics dependence on the input data (e.g., Furthest Point Sampling) to hierarchically select subsets of input points. Thereby, we for the first time propose an end-to-end learnable and task-agnostic sampling operation, named Gumbel Subset Sampling (GSS), to select a representative subset of input points. Equipped with Gumbel-Softmax, it produces a ‘soft’ continuous subset in training phase, and a ‘hard’ discrete subset in test phase. By selecting representative subsets in a hierarchical fashion, the networks learn a stronger representation of the input sets with lower computation cost. Experiments on classification and segmentation benchmarks show the effectiveness and efficiency of our methods. Furthermore, we propose a novel application, to process event camera stream as point clouds, and achieve a state-of-the-art performance on DVS128 Gesture Dataset. …

Autoencoding Variational Transformation (AVT)
The learning of Transformation-Equivariant Representations (TERs), which is introduced by Hinton et al. \cite{hinton2011transforming}, has been considered as a principle to reveal visual structures under various transformations. It contains the celebrated Convolutional Neural Networks (CNNs) as a special case that only equivary to the translations. In contrast, we seek to train TERs for a generic class of transformations and train them in an {\em unsupervised} fashion. To this end, we present a novel principled method by Autoencoding Variational Transformations (AVT), compared with the conventional approach to autoencoding data. Formally, given transformed images, the AVT seeks to train the networks by maximizing the mutual information between the transformations and representations. This ensures the resultant TERs of individual images contain the {\em intrinsic} information about their visual structures that would equivary {\em extricably} under various transformations. Technically, we show that the resultant optimization problem can be efficiently solved by maximizing a variational lower-bound of the mutual information. This variational approach introduces a transformation decoder to approximate the intractable posterior of transformations, resulting in an autoencoding architecture with a pair of the representation encoder and the transformation decoder. Experiments demonstrate the proposed AVT model sets a new record for the performances on unsupervised tasks, greatly closing the performance gap to the supervised models. …

Distill and Transfer Learning (Distral)
Most deep reinforcement learning algorithms are data inefficient in complex and rich environments, limiting their applicability to many scenarios. One direction for improving data efficiency is multitask learning with shared neural network parameters, where efficiency may be improved through transfer across related tasks. In practice, however, this is not usually observed, because gradients from different tasks can interfere negatively, making learning unstable and sometimes even less data efficient. Another issue is the different reward schemes between tasks, which can easily lead to one task dominating the learning of a shared model. We propose a new approach for joint training of multiple tasks, which we refer to as Distral (Distill & transfer learning). Instead of sharing parameters between the different workers, we propose to share a ‘distilled’ policy that captures common behaviour across tasks. Each worker is trained to solve its own task while constrained to stay close to the shared policy, while the shared policy is trained by distillation to be the centroid of all task policies. Both aspects of the learning process are derived by optimizing a joint objective function. We show that our approach supports efficient transfer on complex 3D environments, outperforming several related methods. Moreover, the proposed learning process is more robust and more stable—attributes that are critical in deep reinforcement learning. …

Tuple Plot
Complex systems are described with high-dimensional data that is hard to visualise. Inselberg’s parallel coordinates are one representation technique for visualising high-dimensional data. Here we generalise Inselberg’s approach, and use it for visualising trajectories through high dimensional state spaces. We introduce two geometric projections of parallel coordinate representations — ‘plan tuple plots’ and ‘side tuple plots’ — and demonstrate a link between state space and ordinary space representations. We provide examples from many domains to illustrate use of the approach, including Cellular Automata, Random Boolean Networks, coupled logistic maps, reservoir computing, search algorithms, Turing Machines, and flocking. …

Consumer and Producer Based Recommendation (CPRec)
User-Generated Content (UGC) is at the core of web applications where users can both produce and consume content. This differs from traditional e-Commerce domains where content producers and consumers are usually from two separate groups. In this work, we propose a method CPRec (consumer and producer based recommendation), for recommending content on UGC-based platforms. Specifically, we learn a core embedding for each user and two transformation matrices to project the user’s core embedding into two ‘role’ embeddings (i.e., a producer and consumer role). We model each interaction by the ternary relation between the consumer, the consumed item, and its producer. Empirical studies on two large-scale UGC applications show that our method outperforms standard collaborative filtering methods as well as recent methods that model producer information via item features. …

Backdoor Injection Attack
Deep learning models have consistently outperformed traditional machine learning models in various classification tasks, including image classification. As such, they have become increasingly prevalent in many real world applications including those where security is of great concern. Such popularity, however, may attract attackers to exploit the vulnerabilities of the deployed deep learning models and launch attacks against security-sensitive applications. In this paper, we focus on a specific type of data poisoning attack, which we refer to as a {\em backdoor injection attack}. The main goal of the adversary performing such attack is to generate and inject a backdoor into a deep learning model that can be triggered to recognize certain embedded patterns with a target label of the attacker’s choice. Additionally, a backdoor injection attack should occur in a stealthy manner, without undermining the efficacy of the victim model. Specifically, we propose two approaches for generating a backdoor that is hardly perceptible yet effective in poisoning the model. We consider two attack settings, with backdoor injection carried out either before model training or during model updating. We carry out extensive experimental evaluations under various assumptions on the adversary model, and demonstrate that such attacks can be effective and achieve a high attack success rate (above $90\%$) at a small cost of model accuracy loss (below $1\%$) with a small injection rate (around $1\%$), even under the weakest assumption wherein the adversary has no knowledge either of the original training data or the classifier model. …

Graph Learning Neural Network
Semi-supervised classification on graph-structured data has received increasing attention, where labels are only available for a small subset of data such as social networks and citation networks. This problem is challenging due to the irregularity of graphs. Graph convolutional neural networks (GCN) have been recently proposed to address such kinds of problems, which feed the graph topology into the network to guide operations such as graph convolution. Nevertheless, in most cases where the graphs are not given, they are empirically constructed manually, which tends to be sub-optimal. Hence, we propose Graph Learning Neural Networks (GLNN), which exploits the optimization of graphs (the adjacency matrix in particular) and integrates into the GCN for semi-supervised node classification. Leveraging on spectral graph theory, this essentially combines both graph learning and graph convolution into a unified framework. Specifically, we represent features of social/citation networks as graph signals, and propose the objective of graph learning from the graph-signal prior, sparsity constraint and properties of a valid adjacency matrix via maximum a posteriori estimation. The optimization objective is then integrated into the loss function of the GCN, leading to joint learning of the adjacency matrix and high-level features. Experimental results show that our proposed GLNN outperforms state-of-the-art approaches over widely adopted social network datasets and citation network datasets. …

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. …

Formal Concept Analysis (FCA)
In information science, formal concept analysis is a principled way of deriving a concept hierarchy or formal ontology from a collection of objects and their properties. Each concept in the hierarchy represents the set of objects sharing the same values for a certain set of properties; and each sub-concept in the hierarchy contains a subset of the objects in the concepts above it. The term was introduced by Rudolf Wille in 1984, and builds on applied lattice and order theory that was developed by Garrett Birkhoff and others in the 1930s. Formal concept analysis finds practical application in fields including data mining, text mining, machine learning, knowledge management, semantic web, software development, chemistry and biology. …

TrieJax
Graph pattern matching (e.g., finding all cycles and cliques) has become an important component in many critical domains such as social networks, biology, and cyber-security. This development motivated research to develop faster algorithms that target graph pattern matching. In recent years, the database community has shown that mapping graph pattern matching problems to a new class of relational join algorithms provides an efficient framework for computing these problems. In this paper, we argue that this new class of relational join algorithms is highly amenable to specialized hardware acceleration thanks to two fundamental properties: improved locality and inherent concurrency. The improved locality is a result of the provably bound number of intermediate results these algorithms generate, which results in smaller working sets. In addition, their inherent concurrency can be leveraged for effective hardware acceleration and hiding memory latency. We demonstrate the hardware amenability of this new class of algorithms by introducing TrieJax, a hardware accelerator for graph pattern matching. The TrieJax design leverages the improved locality and high concurrency properties to dramatically accelerate graph pattern matching, and can be tightly integrated into existing manycore processors. We evaluate TrieJax on a set standard graph pattern matching queries and datasets. Our evaluation shows that TrieJax outperforms recently proposed hardware accelerators for graph and database processing that do not employ the new class of algorithms by 7-63x on average (up to 539x), while consuming 15-179x less energy (up to 1750x). systems that do incorporate modern relational join algorithms by 9-20x on average (up to 45x), while consuming 59-110x less energy (up to 372x). …

Self-Adaptive Visual Navigation Method (SAVN)
Learning is an inherently continuous phenomenon. When humans learn a new task there is no explicit distinction between training and inference. After we learn a task, we keep learning about it while performing the task. What we learn and how we learn it varies during different stages of learning. Learning how to learn and adapt is a key property that enables us to generalize effortlessly to new settings. This is in contrast with conventional settings in machine learning where a trained model is frozen during inference. In this paper we study the problem of learning to learn at both training and inference time in the context of visual navigation. A fundamental challenge in navigation is generalization to unseen scenes. In this paper we propose a self-adaptive visual navigation method (SAVN) which learns to adapt to new environments without any explicit supervision. Our solution is a meta-reinforcement learning approach where an agent learns a self-supervised interaction loss that encourages effective navigation. Our experiments, performed in the AI2-THOR framework, show major improvements in both success rate and SPL for visual navigation in novel scenes. …

MatrixRL
Exploration in reinforcement learning (RL) suffers from the curse of dimensionality when the state-action space is large. A common practice is to parameterize the high-dimensional value and policy functions using given features. However existing methods either have no theoretical guarantee or suffer a regret that is exponential in the planning horizon $H$. In this paper, we propose an online RL algorithm, namely the MatrixRL, that leverages ideas from linear bandit to learn a low-dimensional representation of the probability transition model while carefully balancing the exploitation-exploration tradeoff. We show that MatrixRL achieves a regret bound ${O}\big(H^2d\log T\sqrt{T}\big)$ where $d$ is the number of features. MatrixRL has an equivalent kernelized version, which is able to work with an arbitrary kernel Hilbert space without using explicit features. In this case, the kernelized MatrixRL satisfies a regret bound ${O}\big(H^2\widetilde{d}\log T\sqrt{T}\big)$, where $\widetilde{d}$ is the effective dimension of the kernel space. To our best knowledge, for RL using features or kernels, our results are the first regret bounds that are near-optimal in time $T$ and dimension $d$ (or $\widetilde{d}$) and polynomial in the planning horizon $H$. …

Discrete InfoMax Codes (DIMCO)
Learning compact discrete representations of data is itself a key task in addition to facilitating subsequent processing. It is also relevant to meta-learning since a latent representation shared across relevant tasks enables a model to adapt to new tasks quickly. In this paper, we present a method for learning a stochastic encoder that yields discrete p-way codes of length d by maximizing the mutual information between representations and labels. We show that previous loss functions for deep metric learning are approximations to this information-theoretic objective function. Our model, Discrete InfoMax Codes (DIMCO), learns to produce a short representation of data that can be used to classify classes with few labeled examples. Our analysis shows that using shorter codes reduces overfitting in the context of few-shot classification. Experiments show that DIMCO requires less memory (i.e., code length) for performance similar to previous methods and that our method is particularly effective when the training dataset is small. …

Shapley Regression Framework
Machine learning models often excel in the accuracy of their predictions but are opaque due to their non-linear and non-parametric structure. This makes statistical inference challenging and disqualifies them from many applications where model interpretability is crucial. This paper proposes the Shapley regression framework as an approach for statistical inference on non-linear or non-parametric models. Inference is performed based on the Shapley value decomposition of a model, a pay-off concept from cooperative game theory. I show that universal approximators from machine learning are estimation consistent and introduce hypothesis tests for individual variable contributions, model bias and parametric functional forms. The inference properties of state-of-the-art machine learning models – like artificial neural networks, support vector machines and random forests – are investigated using numerical simulations and real-world data. The proposed framework is unique in the sense that it is identical to the conventional case of statistical inference on a linear model if the model is linear in parameters. This makes it a well-motivated extension to more general models and strengthens the case for the use of machine learning to inform decisions. …

Boundary Equilibrium Generative Adversarial Networ (BEGAN)
We propose a new equilibrium enforcing method paired with a loss derived from the Wasserstein distance for training auto-encoder based Generative Adversarial Networks. This method balances the generator and discriminator during training. Additionally, it provides a new approximate convergence measure, fast and stable training and high visual quality. We also derive a way of controlling the trade-off between image diversity and visual quality. We focus on the image generation task, setting a new milestone in visual quality, even at higher resolutions. This is achieved while using a relatively simple model architecture and a standard training procedure.
Tensorflow-BEGAN: Boundary Equilibrium Generative Adversarial Networks …

Memory Augmented Neural Network (MANN)
Neural Turing Machines (NTMs) are an instance of Memory Augmented Neural Networks, a new class of recurrent neural networks which decouple computation from memory by introducing an external memory unit. NTMs have demonstrated superior performance over Long Short-Term Memory Cells in several sequence learning tasks. A number of open source implementations of NTMs exist but are unstable during training and/or fail to replicate the reported performance of NTMs. This paper presents the details of our successful implementation of a NTM. Our implementation learns to solve three sequential learning tasks from the original NTM paper. We find that the choice of memory contents initialization scheme is crucial in successfully implementing a NTM. Networks with memory contents initialized to small constant values converge on average 2 times faster than the next best memory contents initialization scheme. …

Slice Finder
As machine learning (ML) systems become democratized, it becomes increasingly important to help users easily debug their models. However, current data tools are still primitive when it comes to helping users trace model performance problems all the way to the data. We focus on the particular problem of slicing data to identify subsets of the validation data where the model performs poorly. This is an important problem in model validation because the overall model performance can fail to reflect that of the smaller subsets, and slicing allows users to analyze the model performance on a more granular-level. Unlike general techniques (e.g., clustering) that can find arbitrary slices, our goal is to find interpretable slices (which are easier to take action compared to arbitrary subsets) that are problematic and large. We propose Slice Finder, which is an interactive framework for identifying such slices using statistical techniques. Applications include diagnosing model fairness and fraud detection, where identifying slices that are interpretable to humans is crucial. …

DDRL-AM
Learning powerful discriminative features for remote sensing image scene classification is a challenging computer vision problem. In the past, most classification approaches were based on handcrafted features. However, most recent approaches to remote sensing scene classification are based on Convolutional Neural Networks (CNNs). The de facto practice when learning these CNN models is only to use original RGB patches as input with training performed on large amounts of labeled data (ImageNet). In this paper, we show class activation map (CAM) encoded CNN models, codenamed DDRL-AM, trained using original RGB patches and attention map based class information provide complementary information to the standard RGB deep models. To the best of our knowledge, we are the first to investigate attention information encoded CNNs. Additionally, to enhance the discriminability, we further employ a recently developed object function called ‘center loss,’ which has proved to be very useful in face recognition. Finally, our framework provides attention guidance to the model in an end-to-end fashion. Extensive experiments on two benchmark datasets show that our approach matches or exceeds the performance of other methods. …

Short Term Attentive Working Memory Model (STAWM)
The ability to look multiple times through a series of pose-adjusted glimpses is fundamental to human vision. This critical faculty allows us to understand highly complex visual scenes. Short term memory plays an integral role in aggregating the information obtained from these glimpses and informing our interpretation of the scene. Computational models have attempted to address glimpsing and visual attention but have failed to incorporate the notion of memory. We introduce a novel, biologically inspired visual working memory architecture that we term the Hebb-Rosenblatt memory. We subsequently introduce a fully differentiable Short Term Attentive Working Memory model (STAWM) which uses transformational attention to learn a memory over each image it sees. The state of our Hebb-Rosenblatt memory is embedded in STAWM as the weights space of a layer. By projecting different queries through this layer we can obtain goal-oriented latent representations for tasks including classification and visual reconstruction. Our model obtains highly competitive classification performance on MNIST and CIFAR-10. As demonstrated through the CelebA dataset, to perform reconstruction the model learns to make a sequence of updates to a canvas which constitute a parts-based representation. Classification with the self supervised representation obtained from MNIST is shown to be in line with the state of the art models (none of which use a visual attention mechanism). Finally, we show that STAWM can be trained under the dual constraints of classification and reconstruction to provide an interpretable visual sketchpad which helps open the ‘black-box’ of deep learning. …

Stochastic Multicriteria Acceptability Analysis (SMAA)
Stochastic Multicriteria Acceptability Analysis (SMAA) is a multiple-criteria decision analysis method for problems with missing or incomplete information.
A robust hierarchical nominal classification method based on similarity and dissimilarity …

ALAMO
ALAMO is a computational methodology for leaning algebraic functions from data. Given a data set, the approach begins by building a low-complexity, linear model composed of explicit non-linear transformations of the independent variables. Linear combinations of these non-linear transformations allow a linear model to better approximate complex behavior observed in real processes. The model is refined, as additional data are obtained in an adaptive fashion through error maximization sampling using derivative-free optimization. Models built using ALAMO can enforce constraints on the response variables to incorporate first-principles knowledge. The ability of ALAMO to generate simple and accurate models for a number of reaction problems is demonstrated. The error maximization sampling is compared with Latin hypercube designs to demonstrate its sampling efficiency. ALAMO’s constrained regression methodology is used to further refine concentration models, resulting in models that perform better on validation data and satisfy upper and lower bounds placed on model outputs. …

Decentralized Elimination
We consider the \textit {decentralized exploration problem}: a set of players collaborate to identify the best arm by asynchronously interacting with the same stochastic environment. The objective is to insure privacy in the best arm identification problem between asynchronous, collaborative, and thrifty players. In the context of a digital service, we advocate that this decentralized approach allows a good balance between the interests of users and those of service providers: the providers optimize their services, while protecting the privacy of the users and saving resources. We define the privacy level as the amount of information an adversary could infer by intercepting the messages concerning a single user. We provide a generic algorithm {\sc Decentralized Elimination}, which uses any best arm identification algorithm as a subroutine. We prove that this algorithm insures privacy, with a low communication cost, and that in comparison to the lower bound of the best arm identification problem, its sample complexity suffers from a penalty depending on the inverse of the probability of the most frequent players. Finally, we propose an extension of the proposed algorithm to the non-stationary bandits. Experiments illustrate and complete the analysis. …

Graph Similarity Computation via Convolutional Neural Network (GSimCNN)
We introduce GSimCNN (Graph Similarity Computation via Convolutional Neural Networks) for predicting the similarity score between two graphs. As the core operation of graph similarity search, pairwise graph similarity computation is a challenging problem due to the NP-hard nature of computing many graph distance/similarity metrics. We demonstrate our model using the Graph Edit Distance (GED) as the example metric. Experiments on three real graph datasets demonstrate that our model achieves the state-of-the-art performance on graph similarity search. …

ParaNet
DenseNets have been shown to be a competitive model among recent convolutional network architectures. These networks utilize Dense Blocks, which are groups of densely connected layers where the output of a hidden layer is fed in as the input of every other layer following it. In this paper, we aim to improve certain aspects of DenseNet, especially when it comes to practicality. We introduce ParaNet, a new architecture that constructs three pipelines which allow for early inference. We additionally introduce a cascading mechanism such that different pipelines are able to share parameters, as well as logit matching between the outputs of the pipelines. We separately evaluate each of the newly introduced mechanisms of ParaNet, then evaluate our proposed architecture on CIFAR-100. …

POIReviewQ
Many services that perform information retrieval for Points of Interest (POI) utilize a Lucene-based setup with spatial filtering. While this type of system is easy to implement it does not make use of semantics but relies on direct word matches between a query and reviews leading to a loss in both precision and recall. To study the challenging task of semantically enriching POIs from unstructured data in order to support open-domain search and question answering (QA), we introduce a new dataset POIReviewQA. It consists of 20k questions (e.g.’is this restaurant dog friendly?’) for 1022 Yelp business types. For each question we sampled 10 reviews, and annotated each sentence in the reviews whether it answers the question and what the corresponding answer is. To test a system’s ability to understand the text we adopt an information retrieval evaluation by ranking all the review sentences for a question based on the likelihood that they answer this question. We build a Lucene-based baseline model, which achieves 77.0% AUC and 48.8% MAP. A sentence embedding-based model achieves 79.2% AUC and 41.8% MAP, indicating that the dataset presents a challenging problem for future research by the GIR community. The result technology can help exploit the thematic content of web documents and social media for characterisation of locations. …

Budding Perceptron
Traditionally, deep learning algorithms update the network weights whereas the network architecture is chosen manually, using a process of trial and error. In this work, we propose two novel approaches that automatically update the network structure while also learning its weights. The novelty of our approach lies in our parameterization where the depth, or additional complexity, is encapsulated continuously in the parameter space through control parameters that add additional complexity. We propose two methods: In tunnel networks, this selection is done at the level of a hidden unit, and in budding perceptrons, this is done at the level of a network layer; updating this control parameter introduces either another hidden unit or another hidden layer. We show the effectiveness of our methods on the synthetic two-spirals data and on two real data sets of MNIST and MIRFLICKR, where we see that our proposed methods, with the same set of hyperparameters, can correctly adjust the network complexity to the task complexity. …

Multi-Scale Affinity With Sparse Convolution (MASC)
We propose a new approach for 3D instance segmentation based on sparse convolution and point affinity prediction, which indicates the likelihood of two points belonging to the same instance. The proposed network, built upon submanifold sparse convolution [3], processes a voxelized point cloud and predicts semantic scores for each occupied voxel as well as the affinity between neighboring voxels at different scales. A simple yet effective clustering algorithm segments points into instances based on the predicted affinity and the mesh topology. The semantic for each instance is determined by the semantic prediction. Experiments show that our method outperforms the state-of-the-art instance segmentation methods by a large margin on the widely used ScanNet benchmark [2]. We share our code publicly at https://…/MASC. …

Label Augmentation
A major impediment to the application of deep learning to real-world problems is the scarcity of labeled data. Small training sets are in fact of no use to deep networks as, due to the large number of trainable parameters, they will very likely be subject to overfitting phenomena. On the other hand, the increment of the training set size through further manual or semi-automatic labellings can be costly, if not possible at times. Thus, the standard techniques to address this issue are transfer learning and data augmentation, which consists of applying some sort of ‘transformation’ to existing labeled instances to let the training set grow in size. Although this approach works well in applications such as image classification, where it is relatively simple to design suitable transformation operators, it is not obvious how to apply it in more structured scenarios. Motivated by the observation that in virtually all application domains it is easy to obtain unlabeled data, in this paper we take a different perspective and propose a \emph{label augmentation} approach. We start from a small, curated labeled dataset and let the labels propagate through a larger set of unlabeled data using graph transduction techniques. This allows us to naturally use (second-order) similarity information which resides in the data, a source of information which is typically neglected by standard augmentation techniques. In particular, we show that by using known game theoretic transductive processes we can create larger and accurate enough labeled datasets which use results in better trained neural networks. Preliminary experiments are reported which demonstrate a consistent improvement over standard image classification datasets. …

DepecheMood++
Several lexica for sentiment analysis have been developed and made available in the NLP community. While most of these come with word polarity annotations (e.g. positive/negative), attempts at building lexica for finer-grained emotion analysis (e.g. happiness, sadness) have recently attracted significant attention. Such lexica are often exploited as a building block in the process of developing learning models for which emotion recognition is needed, and/or used as baselines to which compare the performance of the models. In this work, we contribute two new resources to the community: a) an extension of an existing and widely used emotion lexicon for English; and b) a novel version of the lexicon targeting Italian. Furthermore, we show how simple techniques can be used, both in supervised and unsupervised experimental settings, to boost performances on datasets and tasks of varying degree of domain-specificity. …

Attribute Manifold Encoding GAN (AME-GAN)
Image attribute transfer aims to change an input image to a target one with expected attributes, which has received significant attention in recent years. However, most of the existing methods lack the ability to de-correlate the target attributes and irrelevant information, i.e., the other attributes and background information, thus often suffering from blurs and artifacts. To address these issues, we propose a novel Attribute Manifold Encoding GAN (AME-GAN) for fully-featured attribute transfer, which can modify and adjust every detail in the images. Specifically, our method divides the input image into image attribute part and image background part on manifolds, which are controlled by attribute latent variables and background latent variables respectively. Through enforcing attribute latent variables to Gaussian distributions and background latent variables to uniform distributions respectively, the attribute transfer procedure becomes controllable and image generation is more photo-realistic. Furthermore, we adopt a conditional multi-scale discriminator to render accurate and high-quality target attribute images. Experimental results on three popular datasets demonstrate the superiority of our proposed method in both performances of the attribute transfer and image generation quality. …

Working Memory Network
During the last years, there has been a lot of interest in achieving some kind of complex reasoning using deep neural networks. To do that, models like Memory Networks (MemNNs) have combined external memory storages and attention mechanisms. These architectures, however, lack of more complex reasoning mechanisms that could allow, for instance, relational reasoning. Relation Networks (RNs), on the other hand, have shown outstanding results in relational reasoning tasks. Unfortunately, their computational cost grows quadratically with the number of memories, something prohibitive for larger problems. To solve these issues, we introduce the Working Memory Network, a MemNN architecture with a novel working memory storage and reasoning module. Our model retains the relational reasoning abilities of the RN while reducing its computational complexity from quadratic to linear. We tested our model on the text QA dataset bAbI and the visual QA dataset NLVR. In the jointly trained bAbI-10k, we set a new state-of-the-art, achieving a mean error of less than 0.5%. Moreover, a simple ensemble of two of our models solves all 20 tasks in the joint version of the benchmark. …