Remaining Useful Life Estimation for Engineered Systems Operating under Uncertainty with Causal GraphNets

In this work, a novel approach, termed GNN-tCNN, is presented for the construction and training of Remaining Useful Life (RUL) models. The method exploits Graph Neural Networks (GNNs) and deals with the problem of efficiently learning from time series with non-equidistant observations, which may span multiple temporal scales. The efficacy of the method is demonstrated on a simulated stochastic degradation dataset and on a real-world accelerated life testing dataset for ball-bearings. The proposed method learns a model that describes the evolution of the system implicitly rather than at the raw observation level and is based on message-passing neural networks, which encode the irregularly sampled causal structure. The proposed approach is compared to a recurrent network with a temporal convolutional feature extractor head (LSTM-tCNN), which forms a viable alternative for the problem considered. Finally, by taking advantage of recent advances in the computation of reparametrization gradients for learning probability distributions, a simple, yet efficient, technique is employed for representing prediction uncertainty as a gamma distribution over RUL predictions.

Machine Learning Technology to Improve Precision and Accelerate Screening Shallow Gas Potentials in Tunu Shallow Gas Zone, PT Pertamina Hulu Mahakam

Tunu is a giant gas field located in the present-day Mahakam Delta, East Kalimantan, Indonesia. Tunu gas produced from Tunu Main Zone (TMZ), between 2500-4500 m TVDSS and Tunu Shallow Zone (TSZ) located on depth 600 - 1500 m TVDSS. Gas reservoirs are scattered along the Tunu Field and corresponds with fluio-deltaic series. Main lithologies are shale, sand, and coal layers. Shallow gas trapping system is a combination of stratigraphic features, and geological structures. The TSZ development relies heavily on the use seismic to assess and identify gas sand reservoirs as drilling targets. The main challenge for conventional use of seismic is differentiating the gas sands from the coal layers. Gas sands are identified by an established seismic workflow that comprises of four different analysis on pre-stack and angle stacks, CDP gathers, amplitude versus angle(AVA), and inversion/litho-seismic cube. This workflow has a high success rate in identifying gas, but requires a lot of time to assess the prospect. The challenge is to assess more than 20,000 shallow objects in TSZ, it is important to have a faster and more efficient workflow to speed up the development phase. The aim of this study is to evaluate the robustness of machine learning to quantify seismic objects/geobodies to be gas reservoirs. We tested various machine learning methods to fit learn geological Tunu characteristic to the seismic data. The training result shows that a gas sand geobody can be predicted using combination of AVA gather, sub-stacks and seismic attributes with model precision of 80%. Two blind wells tests showed precision more than 95% while other final set tests are under evaluated. Detectability here is the ability of machine learning to predicted the actual gas reservoir as compared to the number of gas reservoirs found in that particular wells test. Outcome from this study is expected to accelerate gas assessment workflow in the near future using the machine learning probability cube, with more optimized and quantitative workflow by showing its predictive value in each anomaly.

HFS-SLPEE: A Novel Hierarchical Feature Selection and Second Learning Probability Error Ensemble Model for Precision Cancer Diagnosis

The emergence of high-throughput RNA-seq data has offered unprecedented opportunities for cancer diagnosis. However, capturing biological data with highly nonlinear and complex associations by most existing approaches for cancer diagnosis has been challenging. In this study, we propose a novel hierarchical feature selection and second learning probability error ensemble model (named HFS-SLPEE) for precision cancer diagnosis. Specifically, we first integrated protein-coding gene expression profiles, non-coding RNA expression profiles, and DNA methylation data to provide rich information; afterward, we designed a novel hierarchical feature selection method, which takes the CpG-gene biological associations into account and can select a compact set of superior features; next, we used four individual classifiers with significant differences and apparent complementary to build the heterogeneous classifiers; lastly, we developed a second learning probability error ensemble model called SLPEE to thoroughly learn the new data consisting of classifiers-predicted class probability values and the actual label, further realizing the self-correction of the diagnosis errors. Benchmarking comparisons on TCGA showed that HFS-SLPEE performs better than the state-of-the-art approaches. Moreover, we analyzed in-depth 10 groups of selected features and found several novel HFS-SLPEE-predicted epigenomics and epigenetics biomarkers for breast invasive carcinoma (BRCA) (e.g., TSLP and ADAMTS9-AS2), lung adenocarcinoma (LUAD) (e.g., HBA1 and CTB-43E15.1), and kidney renal clear cell carcinoma (KIRC) (e.g., IRX2 and BMPR1B-AS1).

Machine learning and quantum devices

These brief lecture notes cover the basics of neural networks and deep learning as well as their applications in the quantum domain, for physicists without prior knowledge. In the first part, we describe training using backpropagation, image classification, convolutional networks and autoencoders. The second part is about advanced techniques like reinforce-ment learning (for discovering control strategies), recurrent neural networks (for analyz-ing time traces), and Boltzmann machines (for learning probability distributions). In the third lecture, we discuss first recent applications to quantum physics, with an emphasis on quantum information processing machines. Finally, the fourth lecture is devoted to the promise of using quantum effects to accelerate machine learning.

Jogo de Palitos: Conhecimentos de Estudantes do 5º Ano Sobre Probabilidade em Situação de Jogo

Resumo O presente artigo surgiu a partir do interesse nos jogos que envolvesse Matemática durante o estágio escolar em um projeto chamado Jovens Matemáticos, no qual tinha como propósito a utilização de jogos direcionados a aprendizagem da Matemática. Nesse estudo, optou-se por investigar o uso do jogo de palitos como um suporte para aprendizagem do conteúdo de probabilidade no 5º ano do ensino Fundamental I. Objetivamos analisar a influência do jogo de palitos na compreensão dos estudantes do 5º ano sobre o conteúdo de probabilidade, buscando identificar sua influência, em particular, no que se refere à aleatoriedade, ao espaço amostral e ao cálculo probabilístico. Foi realizado um pré-teste com 18 alunos do 5º Ano para analisar os conhecimentos prévios dos alunos e após a familiarização do jogo realizamos o pós-teste para saber a influência do jogo. Conclui-se que de modo geral o jogo pode ser um instrumento para compreensão do conteúdo em questão nas abordagens do raciocínio probabilísticos, identificação do espaço amostral e estimativas de eventos aleatórios. Palavras-chave: Probabilidade. Jogo dos Palitos. Situação de Jogo. Resume:The present article arose from the interest in games that involved Mathematics during the school stage in a project called Young Mathematicians, whose purpose was the use of games aimed at learning Mathematics. In this study, we chose to investigate the use of matchstick as a support for learning Probability content in the 5th grade of elementary school I. We aimed to analyze the influence of matchstick on the understanding of 5th grade students on the content of Probability, seeking to identify its influence, particularly with regard to randomness, sample space and probabilistic calculation. A pre-test was conducted with 18 5th graders to analyze the students' previous knowledge and after the familiarization of the game we performed the post-test to know the influence of the game. It is concluded that in general the game can be an instrument for understanding the content of Probability in approaches to probabilistic reasoning, sample space identification and random event estimates. Keyword: Probability; Matchsticks; Gambling situation.

PEMBELAJARAN TEORI PROBABILITAS MENGGUNAKAN R

In learning probability theory, if a series of statistical experiments is carried out several times, identifying the possible samples produced is not an easy thing. If this happens, then a significant probability problem will arise. The objective of this study is how to learn probability theory using R software. Based on the simulation results it can be concluded that by combining the syntax in R can be used to solve probability problems such as identifying sample points from experiments, events, event operations, probabilities and conditional probabilities. This will help students learning in understanding of the probability theory material. The use of R will be very pronounced for experiments with a large enough scale that results in a large sample probability.