Clouds for unmanned fireballs and genome analysis
To conduct many studies for universities and scientific
Examples:
The racing team of the Moscow State Technical University N. E.Bauman's (Bauman Racing Team) used cloud power to create an unmanned racing car. The autonomous driving system has an algorithm that recognizes objects on the track in real time. This keeps the car moving in the right direction. For such recognition, the developers used convolutional neural networks, and the training of these neural networks took place in the cloud.
The Bauman Racing Team used the serviceML-development (Machine learning) Yandex DataSphere for training two neural networks that process images. The use of this tool made it possible to significantly increase the speed and convenience of learning algorithms: already now, one of the neural networks has passed 7,000 images through itself, and the second - 3,000.
Center for the Application of Quantitative Methods inBiology at the German University of Tübingen collaborates on its genome research projects with Amazon's cloud service - AWS. German scientists are using the cloud to analyze tens of thousands of genetic samples and identify various patterns, such as differences in gene expression between healthy and diseased tissue. Using the cloud made it possible to reduce the time for researching genomes by 50% and speed up work on the project.
Gene expression- the process of transforming hereditaryinformation (DNA nucleotide sequence) in RNA or protein. It is the regulation of gene expression that gives cells the ability to control their structure. Control of gene expression characteristics influences the functions of other genes throughout the organism.
Count apples, look for cancer, and analyze images of a black hole
Today, cloud providers provide convenient, ready-made tools for ML development and working with data - and this is the second reason why scientific organizations come to use the cloud.
These tools include primarilyservices for machine learning: Yandex DataSphere, Google Cloud ML or Azure ML. Services for easy viewing and visualization of data are of interest to researchers: for example, Yandex DataLens or Microsoft Power BI. There are also more specialized tools such as AWS Panorama for computer vision technologies, Yandex SpeechKit for speech recognition and synthesis, or Google Vision AI for image analysis.
Examples:
Faculty of Biology, Moscow State University, together with the Federal Research Center named afterMichurin, Tambov State University and the VIM agroengineering center used Yandex.Cloud tools to create a garden monitoring system. It is used to more accurately estimate and predict the volume of the harvest and more efficiently plan the further supply chain.
Scientists upload to Yandex cloud storage.Cloud photographs of fruits and use Yandex DataSphere to create a self-learning algorithm: it recognizes individual apples on trees and counts their number, and also learns to determine indicators such as flowering intensity, harvest abundance per hectare and the quality of the fruits themselves.
American Cancer Society basedGoogle Cloud analyzed various images of the organs of women with breast cancer and identified patterns in the occurrence and development of such tumors. Scientists used Cloud ML Engine: machine learning made it possible to carry out analysis 12 times faster. In addition, as the authors of the study note, the use of the cloud provides scientists with the opportunity to scale the results of their work and use the findings in other similar projects.
Международный астрономический исследовательский The Event Horizon Telescope project used the computing power of Google Cloud to create the first image of a black hole. To create such an image, huge amounts of data from telescopes were processed: during the week of observations, an average of 350 TB of data were received daily. To process and analyze such a volume, scientists needed significant power, so they turned to using the cloud.
Protect Baikal, explore artifacts and look for seals
The role of the cloud in the transition of scientific projects is also importantto more modern methods of work: from outdated methods of analysis to the use of artificial intelligence. Everything matters here: the additional capacity available in the cloud, unique services, and the general expertise of cloud providers.
Examples:
One of the most obvious examples of thissynergy - cooperation between Yandex.Cloud and “Point No. 1”, the longest environmental monitoring program in Russia, which has been monitoring the health of Lake Baikal for 75 years. Recently, the project was under threat of closure due to a lack of resources and severely outdated data analysis technologies.
As part of the project, the Yandex.Cloud, together with scientists, is creating an intelligent system for digitally supporting the process of analyzing water samples using AI. To train an algorithm capable of recognizing microorganisms in Baikal water samples, scientists provided more than a thousand images of each type of microorganism. In the future, this “digital assistant” will be able to identify up to 400 species of plankton and recognize up to 99% of samples automatically.
Stanford University took advantage of opportunitiesclouds from AWS to create a database of archaeological finds: we are talking about finds discovered during excavations at the site of Çatalhöyük in Turkey. Previously, creating and regularly updating a database that contained all the information about an item, indicating the exact location of the find and other details about it, required many hours of work by scientists - this took about 20 hours a week, and there was often confusion due to updates entered by different people . Moving the database to the cloud has made the process of updating information much more efficient.
One more example:The US National Oceanic and Atmospheric Administration has used the cloud from Microsoft Azure to create AI tools to help study and protect polar seals and beluga whales in Alaska. Previously, biologists manually studied thousands of aerial photographs and looked for signs of the animals under study, and information during this time often became out of date. Now, AI tools are being used to do the job: training artificial intelligence models and processing the 20 TB of data collected by scientists occurs in the cloud.
Conclude contracts and ensure data security
Sometimes a barrier to using the cloudThere may be a lack of expertise in cloud technologies: the availability of convenient tools and expertise from cloud providers largely solves this problem. Some universities and other government research organizations mistakenly believe that they cannot enter into contracts for the use of cloud technologies, which in fact is not the case.
In some cases, research projects lackfinancial resources. However, many large cloud services have science support programs. In some countries, there are also nationwide projects that should promote the use of the cloud in science. Last year, the US launched the National Research Cloud project, with 22 universities participating, including Stanford University and Carnegie Mellon University, as well as representatives of the US government and Congress and technology corporations including Google, Microsoft, Amazon and IBM. ... The goal of the project is to provide access to cloud capabilities for researchers and find funding for similar projects.
In some cases, scientists may be concernedthe issue of data security in the cloud. In fact, modern cloud platforms have a number of advantages compared to their own physical infrastructure: data is distributed throughout the cloud, and data centers are distributed geographically, so if your data center fails, the cloud concept mitigates this. In addition, the security of the cloud platform is constantly monitored.
The cloud also makes collaboration easieron projects and knowledge sharing: for example, ML tools (DataSphere) allow you to share research results through state saving, so that other scientists can repeat the experiment if necessary.
Science will continue to move to the cloud
All projects whose authors turn to the cloudservices, is united, first of all, by the very fact of using mathematical methods and the associated increase in requirements for the speed and volume of information processing. Scientific institutes come to the cloud primarily for additional computing power, and remain due to the availability of ready-made tools and scalable services.
Read more:
The slowing down of the Earth's rotation caused the release of oxygen on the planet
New particle discovered at the Large Hadron Collider
Scientists have found the oldest example of applied geometry