Dropout is attracting intensive research interest in deep learning as an efficient approach to prevent overfitting. Recently incorporating structural information when deciding which units to drop out produced promising results comparing to methods that ignore the structural information. However, a major issue of the existing work is that it failed to differentiate among instances when constructing the dropout architecture. This can be a significant deficiency for many applications. To solve this issue, we propose Constructivism learning for instance-dependent Dropout Architecture (CODA), which is inspired from a philosophical theory, constructivism learning. Specially, based on the theory we have designed a better drop out technique, Uniform Process Mixture Models, using a Bayesian nonparametric method Uniform process. We have evaluated our proposed method on 5 real-world datasets and compared the performance with other state-of-the-art dropout techniques. The experimental results demonstrated the effectiveness of CODA.

Reinforcement learning is a framework for interactive decision-making with incentives sequentially revealed across time without a system dynamics model. Due to its scaling to continuous spaces, we focus on policy search where one iteratively improves a parameterized policy with stochastic policy gradient (PG) updates. In tabular Markov Decision Problems (MDPs), under persistent exploration and suitable parameterization, global optimality may be obtained. By contrast, in continuous space, the non-convexity poses a pathological challenge as evidenced by existing convergence results being mostly limited to stationarity or arbitrary local extrema. To close this gap, we step towards persistent exploration in continuous space through policy parameterizations defined by distributions of heavier tails defined by tail-index parameter alpha, which increases the likelihood of jumping in state space. Doing so invalidates smoothness conditions of the score function common to PG. Thus, we establish how the convergence rate to stationarity depends on the policy's tail index alpha, a Holder continuity parameter, integrability conditions, and an exploration tolerance parameter introduced here for the first time. Further, we characterize the dependence of the set of local maxima on the tail index through an exit and transition time analysis of a suitably defined Markov chain, identifying that policies associated with Levy Processes of a heavier tail converge to wider peaks. This phenomenon yields improved stability to perturbations in supervised learning, which we corroborate also manifests in improved performance of policy search, especially when myopic and farsighted incentives are misaligned.

Kernel selection plays a central role in determining the performance of Gaussian Process (GP) models, as the chosen kernel determines both the inductive biases and prior support of functions under the GP prior. This work addresses the challenge of constructing custom kernel functions for high-dimensional GP regression models. Drawing inspiration from recent progress in deep learning, we introduce a novel approach named KITT: Kernel Identification Through Transformers. KITT exploits a transformer-based architecture to generate kernel recommendations in under 0.1 seconds, which is several orders of magnitude faster than conventional kernel search algorithms. We train our model using synthetic data generated from priors over a vocabulary of known kernels. By exploiting the nature of the self-attention mechanism, KITT is able to process datasets with inputs of arbitrary dimension. We demonstrate that kernels chosen by KITT yield strong performance over a diverse collection of regression benchmarks.

Neural SDEs combine many of the best qualities of both RNNs and SDEs: memory efficient training, high-capacity function approximation, and strong priors on model space. This makes them a natural choice for modelling many types of temporal dynamics. Training a Neural SDE (either as a VAE or as a GAN) requires backpropagating through an SDE solve. This may be done by solving a backwards-in-time SDE whose solution is the desired parameter gradients. However, this has previously suffered from severe speed and accuracy issues, due to high computational cost and numerical truncation errors. Here, we overcome these issues through several technical innovations. First, we introduce the \textit{reversible Heun method}. This is a new SDE solver that is \textit{algebraically reversible}: eliminating numerical gradient errors, and the first such solver of which we are aware. Moreover it requires half as many function evaluations as comparable solvers, giving up to a $1.98\times$ speedup. Second, we introduce the \textit{Brownian Interval}: a new, fast, memory efficient, and exact way of sampling \textit{and reconstructing} Brownian motion. With this we obtain up to a $10.6\times$ speed improvement over previous techniques, which in contrast are both approximate and relatively slow. Third, when specifically training Neural SDEs as GANs (Kidger et al. 2021), we demonstrate how SDE-GANs may be trained through careful weight clipping and choice of activation function. This reduces computational cost (giving up to a $1.87\times$ speedup) and removes the numerical truncation errors associated with gradient penalty. Altogether, we outperform the state-of-the-art by substantial margins, with respect to training speed, and with respect to classification, prediction, and MMD test metrics. We have contributed implementations of all of our techniques to the torchsde library to help facilitate their adoption.

In this paper, we present a distributed variant of adaptive stochastic gradient method for training deep neural networks in the parameter-server model. To reduce the communication cost among the workers and server, we incorporate two types of quantization schemes, i.e., gradient quantization and weight quantization, into the proposed distributed Adam. Besides, to reduce the bias introduced by quantization operations, we propose an error-feedback technique to compensate for the quantized gradient. Theoretically, in the stochastic nonconvex setting, we show that the distributed adaptive gradient method with gradient quantization and error-feedback converges to the first-order stationary point, and that the distributed adaptive gradient method with weight quantization and error-feedback converges to the point related to the quantized level under both the single-worker and multi-worker modes. At last, we apply the proposed distributed adaptive gradient methods to train deep neural networks. Experimental results demonstrate the efficacy of our methods.

This paper describes the short-term competition on "Components Segmentation Task of Document Photos" that was prepared in the context of the "16th International Conference on Document Analysis and Recognition" (ICDAR 2021). This competition aims to bring together researchers working on the filed of identification document image processing and provides them a suitable benchmark to compare their techniques on the component segmentation task of document images. Three challenge tasks were proposed entailing different segmentation assignments to be performed on a provided dataset. The collected data are from several types of Brazilian ID documents, whose personal information was conveniently replaced. There were 16 participants whose results obtained for some or all the three tasks show different rates for the adopted metrics, like "Dice Similarity Coefficient" ranging from 0.06 to 0.99. Different Deep Learning models were applied by the entrants with diverse strategies to achieve the best results in each of the tasks. Obtained results show that the current applied methods for solving one of the proposed tasks (document boundary detection) are already well stablished. However, for the other two challenge tasks (text zone and handwritten sign detection) research and development of more robust approaches are still required to achieve acceptable results.

With the constant increase of available data in various domains, such as the Internet of Things, Social Networks or Smart Cities, it has become fundamental that agents are able to process and reason with such data in real time. Whereas reasoning over time-annotated data with background knowledge may be challenging, due to the volume and velocity in which such data is being produced, such complex reasoning is necessary in scenarios where agents need to discover potential problems and this cannot be done with simple stream processing techniques. Stream Reasoners aim at bridging this gap between reasoning and stream processing and LASER is such a stream reasoner designed to analyse and perform complex reasoning over streams of data. It is based on LARS, a rule-based logical language extending Answer Set Programming, and it has shown better runtime results than other state-of-the-art stream reasoning systems. Nevertheless, for high levels of data throughput even LASER may be unable to compute answers in a timely fashion. In this paper, we study whether Convolutional and Recurrent Neural Networks, which have shown to be particularly well-suited for time series forecasting and classification, can be trained to approximate reasoning with LASER, so that agents can benefit from their high processing speed.

The amount of information produced, whether by newspapers, blogs and social networks, or by monitoring systems, is increasing rapidly. Processing all this data in real-time, while taking into consideration advanced knowledge about the problem domain, is challenging, but required in scenarios where assessing potential risks in a timely fashion is critical. C-SPARQL, a language for continuous queries over streams of RDF data, is one of the more prominent approaches in stream reasoning that provides such continuous inference capabilities over dynamic data that go beyond mere stream processing. However, it has been shown that, in the presence of huge amounts of data, C-SPARQL may not be able to answer queries in time, in particular when the frequency of incoming data is higher than the time required for reasoning with that data. In this paper, we investigate whether reasoning with C-SPARQL can be approximated using Recurrent Neural Networks and Convolutional Neural Networks, two neural network architectures that have been shown to be well-suited for time series forecasting and time series classification, to leverage on their higher processing speed once the network has been trained. We consider a variety of different kinds of queries and obtain overall positive results with high accuracies while improving processing time often by several orders of magnitude.

Automated source code summarization is a popular software engineering research topic wherein machine translation models are employed to "translate" code snippets into relevant natural language descriptions. Most evaluations of such models are conducted using automatic reference-based metrics. However, given the relatively large semantic gap between programming languages and natural language, we argue that this line of research would benefit from a qualitative investigation into the various error modes of current state-of-the-art models. Therefore, in this work, we perform both a quantitative and qualitative comparison of three recently proposed source code summarization models. In our quantitative evaluation, we compare the models based on the smoothed BLEU-4, METEOR, and ROUGE-L machine translation metrics, and in our qualitative evaluation, we perform a manual open-coding of the most common errors committed by the models when compared to ground truth captions. Our investigation reveals new insights into the relationship between metric-based performance and model prediction errors grounded in an empirically derived error taxonomy that can be used to drive future research efforts

Automated Program Repair (APR) helps improve the efficiency of software development and maintenance. Recent APR techniques use deep learning, particularly the encoder-decoder architecture, to generate patches. Though existing DL-based APR approaches have proposed different encoder architectures, the decoder remains to be the standard one, which generates a sequence of tokens one by one to replace the faulty statement. This decoder has multiple limitations: 1) allowing to generate syntactically incorrect programs, 2) inefficiently representing small edits, and 3) not being able to generate project-specific identifiers. In this paper, we propose Recoder, a syntax-guided edit decoder with placeholder generation. Recoder is novel in multiple aspects: 1) Recoder generates edits rather than modified code, allowing efficient representation of small edits; 2) Recoder is syntax-guided, with the novel provider/decider architecture to ensure the syntactic correctness of the patched program and accurate generation; 3) Recoder generates placeholders that could be instantiated as project-specific identifiers later. We conduct experiments to evaluate Recoder on 395 bugs from Defects4J v1.2, 420 additional bugs from Defects4J v2.0, 297 bugs from IntroClassJava and 40 bugs from QuixBugs. Our results show that Recoder repairs 53 bugs on Defects4J v1.2, which achieves 26.2% (11 bugs) improvement over the previous state-of-the-art approach for single-hunk bugs (TBar). Importantly, to our knowledge, Recoder is the first DL-based APR approach that has outperformed the traditional APR approaches on this benchmark.