Due to lack of data, overfitting ubiquitously exists in real-world applications of deep neural networks (DNNs). In this paper, we propose advanced dropout, a model-free methodology, to mitigate overfitting and improve the performance of DNNs. The advanced dropout technique applies a model-free and easily implemented distribution with a parametric prior, and adaptively adjusts dropout rate. Specifically, the distribution parameters are optimized by stochastic gradient variational Bayes (SGVB) inference in order to carry out an end-to-end training of DNNs. We evaluate the effectiveness of the advanced dropout against nine dropout techniques on five widely used datasets in computer vision. The advanced dropout outperforms all the referred techniques by 0.83% on average for all the datasets. An ablation study is conducted to analyze the effectiveness of each component. Meanwhile, convergence of dropout rate and ability to prevent overfitting are discussed in terms of classification performance. Moreover, we extend the application of the advanced dropout to uncertainty inference and network pruning, and we find that the advanced dropout is superior to the corresponding referred methods. The advanced dropout improves classification accuracies by 4% in uncertainty inference and by 0.2% and 0.5% when pruning more than 90% of nodes and 99.8% of parameters, respectively.

As machine translation (MT) systems progress at a rapid pace, questions of their adequacy linger. In this study we focus on negation, a universal, core property of human language that significantly affects the semantics of an utterance. We investigate whether translating negation is an issue for modern MT systems using 17 translation directions as test bed. Through thorough analysis, we find that indeed the presence of negation can significantly impact downstream quality, in some cases resulting in quality reductions of more than 60%. We also provide a linguistically motivated analysis that directly explains the majority of our findings. We release our annotations and code to replicate our analysis here: https://github.com/mosharafhossain/negation-mt.

This paper introduces a speciation principle for neuroevolution where evolving networks are grouped into species based on the number of hidden neurons, which is indicative of the complexity of the search space. This speciation principle is indivisibly coupled with a novel genotype representation which is characterised by zero genome redundancy, high resilience to bloat, explicit marking of recurrent connections, as well as an efficient and reproducible stack-based evaluation procedure for networks with arbitrary topology. Furthermore, the proposed speciation principle is employed in several techniques designed to promote and preserve diversity within species and in the ecosystem as a whole. The competitive performance of the proposed framework, named Cortex, is demonstrated through experiments. A highly customisable software platform which implements the concepts proposed in this study is also introduced in the hope that it will serve as a useful and reliable tool for experimentation in the field of neuroevolution.

Interpreting how persuasive language influences audiences has implications across many domains like advertising, argumentation, and propaganda. Persuasion relies on more than a message's content. Arranging the order of the message itself (i.e., ordering specific rhetorical strategies) also plays an important role. To examine how strategy orderings contribute to persuasiveness, we first utilize a Variational Autoencoder model to disentangle content and rhetorical strategies in textual requests from a large-scale loan request corpus. We then visualize interplay between content and strategy through an attentional LSTM that predicts the success of textual requests. We find that specific (orderings of) strategies interact uniquely with a request's content to impact success rate, and thus the persuasiveness of a request.

Some of the largest digital consultancies across the globe have come together to assess the state of the global artificial intelligence (AI), revealing that Japanese businesses lead the way when it comes to AI adoption. The study was conducted by US-based research firm ESI ThoughtLab in collaboration with a consortium of digital services and consulting firms operating at the cutting edge of AI. Deloitte, Publicis Sapient, Cognizant, Appen, Dataiku and DataRobot were all involved in the study, which surveyed more than 1,000 companies across 15 countries. The goal was to understand the size and scale of AI initiatives across the world. According to the report, Japan emerges as a surprise leader in AI adoption.

With increasing innovations and adoption across industries, the Voice ID biometrics market is forecast to grow by $2.6 billion in the next 4 years according to analyst firm Technavio. About 32% of the market's growth will originate from APAC during the period. The rise in adoption of biometric voice recognition in the healthcare sector is one of the major factors that will have a positive impact on the growth of this market in the coming years. The rising usage of smartphones and personal digital assistants in the healthcare sector has led to apps for electronic prescriptions, diagnosis and treatment, coding and billing. To prevent unauthorized access to confidential data, these apps could be integrated with voice biometrics leading to greater adoption of voice ID biometrics in the coming years.

The unfulfilled past promises of machine learning in risk, compliance, and information security sectors have been disappointing, though understandable. How on earth do you even begin to look at the mind-boggling labyrinth of tens of thousands of compliance provisions and start threading them together to accelerate efficiency? Now, the change has officially come. So, how did we get to this undeniably exciting point in time – and what does it mean for risk, legal, and compliance professionals? What was only possible theoretically has become a reality thanks to advances in computing power, capacity, cleverly designed software, and cloud computing storage capabilities accessible thanks to clever API.

AI as it's called, is becoming increasingly popular (or unpopular, depending on your view). In 1951, author Arthur C. Clarke published a series of science fiction short stories, including the one he collaborated on 17 years later with movie writer/producer/director Stanley Kubrick, birthing the historic film, "2001: A Space Odyssey." Among other things, Odyssey explored the result of humans interacting with a computer that begins to think like them, and (HAL 9000) takes on a mind of his/its own. When Wozniak and Jobs created Apple, the goal was to get computers to think like man, so they could readily understand each other. That's why the trash icon looks like a garbage receptacle -- "Getting rid of garbage? Throw it in the can."

Pretrained contextualized embeddings are powerful word representations for structured prediction tasks. Recent work found that better word representations can be obtained by concatenating different types of embeddings. However, the selection of embeddings to form the best concatenated representation usually varies depending on the task and the collection of candidate embeddings, and the ever-increasing number of embedding types makes it a more difficult problem. In this paper, we propose Automated Concatenation of Embeddings (ACE) to automate the process of finding better concatenations of embeddings for structured prediction tasks, based on a formulation inspired by recent progress on neural architecture search. Specifically, a controller alternately samples a concatenation of embeddings, according to its current belief of the effectiveness of individual embedding types in consideration for a task, and updates the belief based on a reward. We follow strategies in reinforcement learning to optimize the parameters of the controller and compute the reward based on the accuracy of a task model, which is fed with the sampled concatenation as input and trained on a task dataset. Empirical results on 6 tasks and 23 datasets show that our approach outperforms strong baselines and achieves state-of-the-art performance with fine-tuned embeddings in the vast majority of evaluations.

The process of removing occluding hair has a relevant role in the early and accurate diagnosis of skin cancer. It consists of detecting hairs and restore the texture below them, which is sporadically occluded. In this work, we present a model based on convolutional neural networks for hair removal in dermoscopic images. During the network's training, we use a combined loss function to improve the restoration ability of the proposed model. In order to train the CNN and to quantitatively validate their performance, we simulate the presence of skin hair in hairless images extracted from publicly known datasets such as the PH2, dermquest, dermis, EDRA2002, and the ISIC Data Archive. As far as we know, there is no other hair removal method based on deep learning. Thus, we compare our results with six state-of-the-art algorithms based on traditional computer vision techniques by means of similarity measures that compare the reference hairless image and the one with hair simulated. Finally, a statistical test is used to compare the methods. Both qualitative and quantitative results demonstrate the effectiveness of our network.