Due to the vast amount of reviews available on the Web, in the past decades, a growing share of work has focused on sentiment analysis. Aspect-based sentiment classification is the subtask that seeks to detect the sentiment expressed by the content creators towards a defined target within a sentence. This paper introduces three novel unsupervised attentional neural network models for aspect-based sentiment classification, and tests them on English restaurant reviews. The first model employs an autoencoder-like structure to learn a sentiment embedding matrix where each row of the matrix represents the embedding for one sentiment. To improve the model, a target-based attention mechanism is included that de-emphasizes irrelevant words. Last, a redundancy and a seed regularization term constrain the sentiment embedding matrix. The second model extends the first by including a Bi-LSTM layer in the attention mechanism to exploit contextual information. The third model further adapts a Left-Center-Right separated neural network with Rotatory attention structure from the supervised realm to an unsupervised setting. Although all three models construct meaningful sentiment embeddings, experimental results indicate that the inclusion of the Bi-LSTM in the attention mechanism leads to a more precise attention mechanism and, thus, better predictions. The best model, i.e., the second, outperforms all investigated unsupervised and weakly supervised algorithms for aspect-based sentiment classification from the literature.

Fog Computing is a paradigm decentralizing the Cloud by geographically distributing computation, storage, network resources and related services. It provides benefits such as reducing the number of bottlenecks, limiting unwanted data movements, etc. However, managing the size, complexity and heterogeneity of the Fog systems to be engineered is challenging and can quickly become costly. According to best practices in software engineering, verification tasks could be performed on a system design prior to its implementation and deployment. We propose a generic model-based approach for verifying Fog systems at design time, also enabling the automatic generation of corresponding deployment configuration files. Named VeriFog, this approach is notably based on a customizable Fog Modeling Language (FML). We experimented in practice by modeling three use cases, from three different application domains, and by considering three main types of non-functional properties to be verified. From this modeling and verification effort, we show that we are able to automatically generate usable deployment configuration files for different deployment tools. In direct collaboration with our industrial partner Smile, the approach and underlying language presented in this paper are necessary steps towards a more global model-based support for the complete life cycle of Fog systems.

Concrete domains have been introduced in description logic (DL) to enable reference to concrete objects (such as numbers) and predefined predicates on these objects (such as numerical comparisons) when defining concepts. The primary research goal in this context was to find restrictions on the concrete domain such that its integration into certain DLs preserves decidability or tractability. In this article, we complement these investigations by studying the abstract expressive power of both first-order and description logics extended with concrete domains, i.e., we analyze which classes of first-order interpretations can be expressed using these logics, compared to what their counterparts without concrete domains can express. We demonstrate that, under natural conditions on the concrete domain D (which also play a role for decidability), extensions of first-order logic (FOL) or the well-known DL ALC with D share important formal characteristics with FOL, such as the compactness and the Löwenheim-Skolem properties. Nevertheless, these conditions do not ensure that the abstract expressive power of the extensions we consider is contained in that of FOL, though in some cases it is. To be more precise, we show, on the one hand, that unary concrete domains leave the abstract expressive power within FOL if we are allowed to introduce auxiliary predicates. As a by-product, we obtain (semi-)decidability results for some fragments of FOL extended with the concrete domains considered in this article. On the other hand, we show that the ability to express equality between elements of D , another condition employed in the context of showing decidability of ALC ( D ), is sufficient to push the abstract expressive power of most first-order fragments with concrete domains beyond that of FOL. While for such concrete domains D decidability is retained for ALC ( D ), we show that the availability of equality in D causes undecidability of the two-variable fragment of FOL( D ), although the two-variable fragment of FOL is decidable.

In this paper we present C o R o NN a a deep sequential framework for epidemic prediction that leverages a flexible combination of sequential and convolutional components to analyse the transmission of COVID-19 and, perhaps, other undiscovered viruses. Importantly, our methodology is generic and may be tailored to specific analysis goals. We exploit C o R o NN a to analyse the impact of various mobility containment policies on the pandemic using cumulative viral dissemination statistics with local demographic and movement data from several nations. Our experiments show that data on mobility has a significant, but delayed, impact on viral propagation. When compared to alternative frameworks that combine multivariate lagged predictors and basic LSTM models, C o R o NN a outperforms them. On the contrary, no technique based solely on lagged viral dissemination statistics can forecast daily cases.

In this paper, we provide a fine-grain machine learning-based method, PerfNetV2, which improves the accuracy of our previous work for modeling the neural network performance on a variety of GPU accelerators. Given an application, the proposed method can be used to predict the inference time and training time of the convolutional neural networks used in the application, which enables the system developer to optimize the performance by choosing the neural networks and/or incorporating the hardware accelerators to deliver satisfactory results in time. Furthermore, the proposed method is capable of predicting the performance of an unseen or non-existing device, e.g. a new GPU which has a higher operating frequency with less processor cores, but more memory capacity. This allows a system developer to quickly search the hardware design space and/or fine-tune the system configuration. Compared to the previous works, PerfNetV2 delivers more accurate results by modeling detailed host-accelerator interactions in executing the full neural networks and improving the architecture of the machine learning model used in the predictor. Our case studies show that PerfNetV2 yields a mean absolute percentage error within 13.1% on LeNet, AlexNet, and VGG16 on NVIDIA GTX-1080Ti, while the error rate on a previous work published in ICBD 2018 could be as large as 200%.

Modern containerized distributed systems, such as big data storage and processing stacks or micro-service based applications, are inherently hard to monitor and optimize, as resource usage does not directly match hardware resources due to multiple virtualization layers. For instance, interapplication traffic is an important factor in as it directly indicates how components interact, it has not been possible to accurately monitor it in an application independent way and without severe overhead, thus putting it out of reach of cloud platforms. In this paper we present an efficient black-box monitoring approach for gathering detailed structural information of collaborating processes in a distributed system that can be queried for various purposes, as it includes both information about processes, containers, and hosts, as well as resource usage and amount of data exchanged. The key to achieving high detail and low overhead without custom application instrumentation is to use a kernel-aided event driven strategy. We validate a prototype implementation by applying it to multi-platform microservice deployments, evaluate its performance with micro-benchmarks, and demonstrate its usefulness for container placement in a distributed data storage and processing stack (i.e., Cassandra and Spark).

Accurate prediction of the future position of pedestrians in traffic scenarios is required for safe navigation of an autonomous vehicle but remains a challenge. This concerns, in particular, the effective and efficient multimodal prediction of most likely trajectories of tracked pedestrians from egocentric view of self-driving car. In this paper, we present a novel solution, named M2P3, which combines a conditional variational autoencoder with recurrent neural network encoder-decoder architecture in order to predict a set of possible future locations of each pedestrian in a traffic scene. The M2P3 system uses a sequence of RGB images delivered through an internal vehicle-mounted camera for egocentric vision. It takes as an input only two modes, that are past trajectories and scales of pedestrians, and delivers as an output the three most likely paths for each tracked pedestrian. Experimental evaluation of the proposed architecture on the JAAD, ETH/UCY and Stanford Drone datasets reveal that the M2P3 system is significantly superior to selected state-of-the-art solutions.

Most of the online queries target entities and the type of an entity is a key piece of information. Entity type helps us to understand what an entity is and how it relates to other entities mentioned in a document. Search engine result pages (SERPs) often surface facts and entity type information from a background Knowledge Graph (KG) in response to queries that carry a semantic information need. In a KG, an entity usually holds multiple type properties. For example, popular types attached to the entity `Donald Trump' via rdfs:type statements might be Person, Businessman, and Leader. However, other types of this entity, e.g., Solicitor, Restaurateur, and Writer might also be interesting to some users. Unpopular entity types can be useful for tail queries like, for example, `Is Donald Trump an American Television Producer' or `Is Donald Trump an American Television Actor'. It is then important to, given an entity in a KG, rank entity types attached to the entity by relevance to a certain user and information need as not always the most popular type is the most informative within a textual context. In this paper we address the entity type ranking problem by means of KG embedding models. In our work, we show that entity type ranking can be seen as a special case of the KG completion problem. Embeddings can be learned from the structural, probabilistic and contextual description information of the entities. We propose and evaluate our methods to find the most relevant entity type based on collection statistics and on the graph structure interconnecting entities and types. Experimental results show that our proposed approaches outperform the state-of-the-art type ranking models while, at the same time, being more efficient and scalable. Our approach focuses on the task of ranking a set of types associated to an entity in a background knowledge graph to select the most relevant types.

Serverless computing through Function-as-a-Service (FaaS) platforms is a popular new cloud computing model, offering high elasticity while being fully-managed. However, FaaS platforms do not quantify elasticity; thus, application developers are unaware of how elastic FaaS platforms are. This paper is an extended version of [18]. The original paper proposes an experiment design and corresponding toolkit for observing elasticity and its associated trade-offs with latency, reliability, and execution costs. Results for evaluating four popular FaaS platforms by AWS, Google, IBM, and Microsoft indicate significant short-comings in terms of elasticity for all but AWS's service offer. One year later, we reproduced all experiments and revisited our discussions. We report significant improvements by Google and IBM. Our results indicate that (currently) all but Microsoft's FaaS platform are elastic with differences in latency and execution costs.

Genetic programming (GP) is an artificial intelligence technique that benefits from evolutionary computations allowing computers to solve problems automatically. In this paper, we present an optimized genetic-programming-based classifier that directly solves the multi-class classification problems in data mining and image analysis. A new fitness function is proposed for multiclass classification and brain tumor detection, which is validated by 10-fold cross validation. Instead of defining static thresholds as boundaries to differentiate between multiple labels, our work presents a method of classification where a GP system learns the relationships among experiential data and models them mathematically during the evolutionary process. We propose an optimized GP classifier based on a combination of pruning subtrees and a new fitness function. An orthogonal least squares algorithm is also applied in the training phase to create a robust GP classifier. Our approach has been assessed on six multiclass datasets and on a magnetic resonance imaging (MRI) brain image for tumor detection. The results of data classification for Iris, Wine, Glass, Pima, BUPA Liver and Balance Scale datasets are compared with existing algorithms. The high accuracy of brain tumor classification provided by our GP classifier confirms the strong ability of the developed technique for complicated classification problems. We compared our approach in terms of speed with previous GP algorithms as well. The analyzed results illustrate that the developed classifier produces a productive and rapid method for classification tasks that outperforms the previous methods for more challenging multiclass classification problems.