Search

Machine learning is the study of computer algorithms that focuses on analyzing and interpreting patterns and structures in data. It has been successfully applied to many areas in computer science and achieved state-of-the-art results to enable learning, reasoning, and decision-making without human interactions. This research aims to develop innovated data parallel frameworks to accommodate the computing resources to parallelize different machine learning and deep learning algorithms and speed up the training. To achieve that, we explore three interesting frameworks in this dissertation: (1) Sync-on-the-fly framework for gradient descent algorithms on transient resources; (2) Asynchronous Proactive Data Parallel framework for both gradient descent and Expectation-Maximization algorithms; (3) Cohesive Mini-batches graph convolutional network framework for graph convolutional networks.

Forecast of electricity is one of the most important topics in power system, it means a lot for power supply unit that forecast can be more accurate. Traditional methods always use single algorithm which has certain limitations. This article used two predict methods—back propagation (BP) neural network and combination of Auto-regressive Integrated Moving Average Model (ARIMA) and Support Vector Regression (SVR), considering macroeconomic indicators, yield of upstream and downstream products, weathers and other indicators to predict twenty energy-intensive enterprises’ electricity consumption. As a consequence, ARIMA+SVR performs better in predicting.

The purpose of this article is to solve the safety and marketing problems of milk powder, let consumers know more product information, enable consumers to buy at ease, and reduce the burden of manufacturers’ sales. By tracking the whole process of milk powder production, processing, packaging, and transportation, the traceability of the problem products can be achieved. At the same time, the manufacturers can also reasonably make sales plan to the milk powder sales through the rationalization sales system. To ensure the quality and safety of milk powder, ensure that problems in the production process are solved in time and prevent potential harm to infants. Through the collection and recording of data and information, the contradiction between supply and demand of milk powder is effectively solved, and the traceability and rationalized sales system of milk powder are realized.

Summary The cyclic adenosine monophosphate–protein kinase A (cAMP–PKA) pathway is a central signalling cascade that transmits extracellular stimuli and governs cell responses through the second messenger cAMP. The importance of cAMP signalling in fungal biology has been well documented and the key conserved components, adenylate cyclase (AC) and the catalytic subunit of PKA (CPKA), have been functionally characterized. However, other genes involved in this signalling pathway and their regulation are not well understood in filamentous fungi. Here, we performed a comparative transcriptomics analysis of AC and CPKA mutants in two closely related fungi: Fusarium graminearum (Fg) and F. verticillioides (Fv). Combining available Fg transcriptomics and phenomics data, we reconstructed the Fg cAMP signalling pathway. We developed a computational program that combines sequence conservation and patterns of orthologous gene expression to facilitate global transcriptomics comparisons between different organisms. We observed highly correlated expression patterns for most orthologues (80%) between Fg and Fv. We also identified a subset of 482 (6%) diverged orthologues, whose expression under all conditions was at least 50% higher in one genome than in the other. This enabled us to dissect the conserved and unique portions of the cAMP–PKA pathway. Although the conserved portions controlled essential functions, such as metabolism, the cell cycle, chromatin remodelling and the oxidative stress response, the diverged portions had species‐specific roles, such as the production and detoxification of secondary metabolites unique to each species. The evolution of the cAMP–PKA signalling pathway seems to have contributed directly to fungal divergence and niche adaptation.