Most of the information is stored as text, so text mining is regarded as having high commercial potential. Aiming at the semant ic constraint problem of classification methods based on sparse representation, we propose a weighted recurrent neural network (W - RNN), which can fully extract text serialization semantic information. For the problem that the feature high dimensionality an d unclear semantic relationship in text data representation, we first utilize the word vector to represent the vocabulary in the text and use Recurrent Neural Network (RNN) to extract features of the serialized text data. The word vector is then automatica lly weighted and summed using the intermediate output of the word vector to form the text representation vector. Finally, the neural network is used for classification. W - RNN is verified on the news dataset and proves that W - RNN is superior to other four b aseline methods in Precision, Recall, F1 and loss values, which is suitable for text classification. On account of the certainty and comprehensibility of its expression, text has bec ome the popular way of information expression and transmission. Text classification is an extremely important research direction [1].

Combinatorial optimization is the field devoted to the study and practice of algorithms that solve NP-hard problems. As Machine Learning (ML) and deep learning have popularized, several research groups have started to use ML to solve combinatorial optimization problems, such as the well-known Travelling Salesman Problem (TSP). Based on deep (reinforcement) learning, new models and architecture for the TSP have been successively developed and have gained increasing performances. At the time of writing, state-of-the-art models provide solutions to TSP instances of 100 cities that are roughly 1.33% away from optimal solutions. However, despite these apparently positive results, the performances remain far from those that can be achieved using a specialized search procedure. In this paper, we address the limitations of ML approaches for solving the TSP and investigate two fundamental questions: (1) how can we measure the level of accuracy of the pure ML component of such methods; and (2) what is the impact of a search procedure plugged inside a ML model on the performances? To answer these questions, we propose a new metric, ratio of optimal decisions (ROD), based on a fair comparison with a parametrized oracle, mimicking a ML model with a controlled accuracy. All the experiments are carried out on four state-of-the-art ML approaches dedicated to solve the TSP. Finally, we made ROD open-source in order to ease future research in the field.

The dependency of the generalization error of neural networks on model and dataset size is of critical importance both in practice and for understanding the theory of neural networks. Nevertheless, the functional form of this dependency remains elusive. In this work, we present a functional form which approximates well the generalization error in practice. Capitalizing on the successful concept of model scaling (e.g., width, depth), we are able to simultaneously construct such a form and specify the exact models which can attain it across model/data scales. Our construction follows insights obtained from observations conducted over a range of model/data scales, in various model types and datasets, in vision and language tasks. We show that the form both fits the observations well across scales, and provides accurate predictions from small- to large-scale models and data.

Accurate forward modeling is important for solving inverse problems. An inaccurate wave-equation simulation, as a forward operator, will offset the results obtained via inversion. In this work, we consider the case where we deal with incomplete physics. One proxy of incomplete physics is an inaccurate discretization of Laplacian in simulation of wave equation via finite-difference method. We exploit intrinsic one-to-one similarities between timestepping algorithm with Convolutional Neural Networks (CNNs), and propose to intersperse CNNs between low-fidelity timesteps. Augmenting neural networks with low-fidelity timestepping algorithms may allow us to take large timesteps while limiting the numerical dispersion artifacts. While simulating the wave-equation with low-fidelity timestepping algorithm, by correcting the wavefield several time during propagation, we hope to limit the numerical dispersion artifact introduced by a poor discretization of the Laplacian. As a proof of concept, we demonstrate this principle by correcting for numerical dispersion by keeping the velocity model fixed, and varying the source locations to generate training and testing pairs for our supervised learning algorithm.

Recently, how to expand data transmission to reduce cell data and repeated cell transmission has received more and more research attention. In mobile social networks, content popularity prediction has always been an important part of traffic offloading and expanding data dissemination. However, current mainstream content popularity prediction methods only use the number of downloads and shares or the distribution of user interests, which do not consider important time and geographic location information in mobile social networks, and all of data is from OSN which is not same as MSN. In this work, we propose D2D Long Short-Term Memory (D2D-LSTM), a deep neural network based on LSTM, which is designed to predict a complete D2D diffusion path. Our work is the first attempt in the world to use real data of MSN to predict diffusion path with deep neural networks which conforms to the D2D structure. Compared to linear sequence networks, only learn users' social features without time distribution or GPS distribution and files' content features, our model can predict the propagation path more accurately (up to 85.858\%) and can reach convergence faster (less than 100 steps) because of the neural network that conforms to the D2D structure and combines user social features and files features. Moreover, we can simulate generating a D2D propagation tree. After experiment and comparison, it is found to be very similar to the ground-truth trees. Finally, we define a user prototype refinement that can more accurately describe the propagation sharing habits of a prototype user (including content preferences, time preferences, and geographic location preferences), and experimentally validate the predictions when the user prototype is added to 1000 classes, it is almost identical to the 50 categories.

We present Generative Adversarial rePresentations (GAP) as a data-driven framework for learning censored and/or fair representations. GAP leverages recent advancements in adversarial learning to allow a data holder to learn universal representations that decouple a set of sensitive attributes from the rest of the dataset. Under GAP, finding the optimal mechanism? {decorrelating encoder/decorrelator} is formulated as a constrained minimax game between a data encoder and an adversary. We show that for appropriately chosen adversarial loss functions, GAP provides {censoring} guarantees against strong information-theoretic adversaries and enforces demographic parity. We also evaluate the performance of GAP on multi-dimensional Gaussian mixture models and real datasets, and show how a designer can certify that representations learned under an adversary with a fixed architecture perform well against more complex adversaries.

Modeling sequential data has become more and more important in practice. Some applications are autonomous driving, virtual sensors and weather forecasting. To model such systems, so called recurrent models are frequently used. In this paper we introduce several new Deep recurrent Gaussian process (DRGP) models based on the Sparse Spectrum Gaussian process (SSGP) and the improved version, called variational Sparse Spectrum Gaussian process (VSSGP). We follow the recurrent structure given by an existing DRGP based on a specific variational sparse Nystr\"om approximation, the recurrent Gaussian process (RGP). Similar to previous work, we also variationally integrate out the input-space and hence can propagate uncertainty through the Gaussian process (GP) layers. Our approach can deal with a larger class of covariance functions than the RGP, because its spectral nature allows variational integration in all stationary cases. Furthermore, we combine the (variational) Sparse Spectrum ((V)SS) approximations with a well known inducing-input regularization framework. We improve over current state of the art methods in prediction accuracy for experimental data-sets used for their evaluation and introduce a new data-set for engine control, named Emission.

Deep neural networks (DNNs) are known for extracting good representations from a large amount of data. However, the representations learned in DNNs are typically hard to interpret, especially the ones learned in dense layers. One crucial issue is that neurons within each layer of DNNs are conditionally independent with each other, which makes the co-training and analysis of neurons at higher modularity difficult. In contrast, the dependency patterns of biological neurons in the human brain are largely different from those of DNNs. Neuronal assembly describes such neuron dependencies that could be found among a group of biological neurons as having strong internal synaptic interactions, potentially high semantical correlations that are deemed to facilitate the memorization process. In this paper, we show such a crucial gap between DNNs and biological neural networks (BNNs) can be bridged by the newly proposed Biologically-Enhanced Artificial Neuronal assembly (BEAN) regularization that could enforce dependencies among neurons in dense layers of DNNs without altering the conventional architecture. Both qualitative and quantitative analyses show that BEAN enables the formations of interpretable and biologically plausible neuronal assemblies in dense layers and consequently enhances the modularity and interpretability of the hidden representations learned. Moreover, BEAN further results in sparse and structured connectivity and parameter sharing among neurons, which substantially improves the efficiency and generalizability of the model.

Machine learning as a service (MLaaS), and algorithm marketplaces are on a rise. Data holders can easily train complex models on their data using third party provided learning codes. Training accurate ML models requires massive labeled data and advanced learning algorithms. The resulting models are considered as intellectual property of the model owners and their copyright should be protected. Also, MLaaS needs to be trusted not to embed secret information about the training data into the model, such that it could be later retrieved when the model is deployed. In this paper, we present \emph{membership encoding} for training deep neural networks and encoding the membership information, i.e. whether a data point is used for training, for a subset of training data. Membership encoding has several applications in different scenarios, including robust watermarking for model copyright protection, and also the risk analysis of stealthy data embedding privacy attacks. Our encoding algorithm can determine the membership of significantly redacted data points, and is also robust to model compression and fine-tuning. It also enables encoding a significant fraction of the training set, with negligible drop in the model's prediction accuracy.

T HE D IFFERENTIABLEC ROSS-E NTROPYM ETHOD Brandon Amos 1 Denis Y arats 12 1 Facebook AI Research 2 New Y ork University A BSTRACT We study the Cross-Entropy Method (CEM) for the non-convex optimization of a continuous and parameterized objective function and introduce a differentiable variant (DCEM) that enables us to differentiate the output of CEM with respect to the objective function's parameters. In the machine learning setting this brings CEM inside of the end-to-end learning pipeline where this has otherwise been impossible. We show applications in a synthetic energy-based structured prediction task and in non-convex continuous control. In this paper we focus on the setting of optimizing an unconstrained, non-convex, and continuous objective function f θ(x): R n Θ R as ˆ x arg min x f θ(x), where f is parameterized by θ Θ and has inputs x R n . If it exists, some (sub-)derivative θˆ x is useful in the machine learning setting to make the output of the optimization procedure end-to-end learnable. For example, θ could parameterize a predictive model that is generating potential outcomes conditional on x happening that you want to optimize over. End-to-end learning in these settings can be done by defining a loss function L on top of ˆ x and taking gradient steps θL . If f θ were convex this gradient is easy to analyze and compute when it exists and is unique (Gould et al., 2016; Johnson et al., 2016; Amos et al., 2017; Amos & Kolter, 2017). Unfortunately analyzing and computing a "derivative" through the non-convex arg min here is not as easy and is challenging in theory and practice. No such derivative may exist in theory, it might not be unique, and even if it uniquely exists, the numerical solver being used to compute the solution may not find a global or even local optimum of f . One promising direction to sidestep these issues is to approximate the arg min operation with an explicit optimization procedure that is interpreted as just another compute graph and unrolled through.