Neural networks that analyse electrocardiograms can be easily fooled, mistaking your normal heartbeat reading as irregular or vice versa, researchers warn in a paper published in Nature Medicine.

ECG sensors are becoming more widespread, embedded in wearable devices like smartwatches, while machine learning software is being increasingly developed to automatically monitor and process data to tell users about their heartbeats. The US Food and Drug Administration approved 23 algorithms for medical use in 2018 alone.

However, the technology isnt foolproof. Like all deep learning models, ECG ones are susceptible to adversarial attacks: miscreants can force algorithms to misclassify the data by manipulating it with noise.

A group of researchers led by New York University demonstrated this by tampering with a deep convolutional neural network (CNN). First, they obtained a dataset containing 8,528 ECG recordings labelled into four groups: Normal, atrial fibrillation - the most common type of an irregular heartbeat - other, or noise.

The majority of the dataset, some 5,076 samples were considered normal, 758 fell into the atrial fibrillation category, 2,415 classified as other, and 279 as noise. The researchers split the dataset and used 90 per cent of it to train the CNN, and the other 10 per cent to test the system.

Deep learning classifiers are susceptible to adversarial examples, which are created from raw data to fool the classifier such that it assigns the example to the wrong class, but which are undetectable to the human eye, the researchers explained in the paper (Here's the free preprint version of the paper on arXiv.)

To create these adversarial examples, the researchers added a small amount of noise to samples used in the test set. The uniform peaks and troughs in ECG reading may appear innocuous and normal to the human eye, but adding a small interference was enough to trick the CNN into classifying them as atrial fibrillation - an irregular heartbeat linked to heart palpitations and an increased risk of strokes.

Here are two adversarial examples. The first one shows how an irregular atrial fibrillation (AF) reading being misclassified as normal. The second one is a normal reading misclassified as irregular. Image Credit: Tian et al. and Nature Medicine.

When the researchers fed the adversarial examples to the CNN, 74 per cent of the readings that were originally correctly classified were subsequently wrong. In other words, the model mistook 74 per cent of the readings by assigning them to incorrect labels. What was originally a normal reading then seemed irregular, and vice versa.

Luckily, humans are much more difficult to trick. Two clinicians were given pairs of readings - an original, unperturbed sample and its corresponding adversarial example and asked if either of them looked like they belonged to a different class. They only thought 1.4 per cent of the readings should have been labelled differently.

The heartbeat patterns in original and adversarial samples looked similar to the human eye, and, therefore, itd be fairly easy to tell if a normal heartbeat had been incorrectly misclassified as irregular. In fact, both experts were able to tell the original reading from the adversarial one about 62 per cent of the time.

The ability to create adversarial examples is an important issue, with future implications including robustness to the environmental noise of medical devices that rely on ECG interpretation - for example, pacemakers and defibrillators - the skewing of data to alter insurance claims and the introduction of intentional bias into clinical trial, the paper said.

Its unclear how realistic these adversarial attacks truly are in the real world, however. In these experiments, the researchers had full access to the model making it easy to attack but its much more difficult for these types of attacks to work on, say, someones Apple Watch, for example.

The Register has contacted the researchers for comment. But what the research does prove, however, is that relying solely on machines may be unreliable and that specialists really ought to double check results when neural networks are used in clinical settings.

In conclusion, with this work, we do not intend to cast a shadow on the utility of deep learning for ECG analysis, which undoubtedly will be useful to handle the volumes of physiological signals requiring processing in the near future, the researchers wrote.

This work should, instead, serve as an additional reminder that machine learning systems deployed in the wild should be designed with safety and reliability in mind, with a particular focus on training data curation and provable guarantees on performance.

Sponsored: Quit your addiction to storage

Read more:
Think your smartwatch is good for warning of a heart attack? Turns out it's surprisingly easy to fool its AI - The Register

What You Need to Know:

New Chilmark Research report reveals artificial intelligence and machine learning (AI/ML) technologies are capturing the imagination of investors and healthcare organizationsand are poised to expand healthcare frontiers.

The latest report evaluates over 120 commercial AI/ML solutions in healthcare, explores future opportunities, and assesses obstacles to adoption at scale.

Interest and investment in healthcare AI/ML toolsis booming with approximately $4B in capital funding pouring into thishealthcare sector in 2019. Such investment is spurring a vast array of AI/MLtools for providers, patients, and payers accelerating the possibilities fornew solutions to improve diagnostic accuracy, improve feedback mechanisms, andreduce clinical and administrative errors, according to Chilmark Researchs last report.

The Promise of AI & ML in Healthcare ReportBackground

The report,The Promise of AI & ML in Healthcare, is the most comprehensive report published on this rapidly evolving market with nearly 120 vendors profiled. The report explores opportunities, trends, and the rapidly evolving landscape for vendors, tracing the evolution from early AI/ML use in medical imaging to todays rich array of vendor solutions in medical imaging, business operations, clinical decision support, research and drug development, patient-facing applications, and more. The report also reviews types and applications of AI/ML, explores the substantial challenges of health data collection and use, and considers issues of bias in algorithms, ethical and governance considerations, cybersecurity, and broader implications for business.

Health IT vendors, new start-up ventures, providers, payers,and pharma firms now offer (or are developing) a wide range of solutions for anequally wide range of industry challenges. Our extensive research for thisreport found that nearly 120 companies now offer AI-based healthcare solutionsin four main categories: hospital operations, clinical support, research anddrug development, and patient/consumer engagement.

Report Key Themes

This report features an overview of these major areas of AI/ML use in healthcare. Solutions for hospital operations include tools for revenue cycle management, applications to detect fraud detection and ensure payment integrity, administrative and supply chain applications to improve hospital operations, and algorithms to boost patient safety. Population health management is an area ripe in AI/ML innovation, with predictive analytics solutions devoted to risk stratification, care management, and patient engagement.

A significant development is underway in AI/ML solutions for clinical decision support, including NLP- and voice-enabled clinical documentation applications, sophisticated AI-based medical imaging and pathology tools, and electronic health records management tools to mitigate provider burnout. AI/ML-enabled tools are optimizing research and drug development by improving clinical trials and patient monitoring, modeling drug simulations, and enabling precision medicine advancement. A wealth of consumer-facing AI/ML applications, such as chatbots, wearables, and symptom checkers, are available and in development.

Provider organizations will find this report offers deep insight into current and forthcoming solutions that can help support business operations, population health management, and clinical decision support. Current and prospective vendors of AI/ML solutions and their investors will find this reports overview of the current market valuable in mapping their own product strategy. Researchers and drug developers will benefit from the discussion of current AI/ML applications and future possibilities in precision medicine, clinical trials, drug discovery, and basic research. Providers and patient advocates will gain valuable insight into patient-facing tools currently available and in development.

All stakeholders in healthcare technologyproviders, payers, pharmaceutical stakeholders, consultants, investors, patient advocates, and government representativeswill benefit from a thorough overview of current offerings as well as thoughtful discussions of bias in data collection and underlying algorithms, cyber-security, governance, and ethical concerns.

For more information about the report, please visit https://www.chilmarkresearch.com/chilmark_report/the-promise-of-ai-and-ml-in-healthcare-opportunities-challenges-and-vendor-landscape/

Read this article:
Chilmark Research: The Promise of AI & ML in Healthcare Report - HIT Consultant

Yunkai Sun,1,2,* Lei Zhao,1,2,* Zhihui Lan,1,2 Xi-Ze Jia,1,2 Shao-Wei Xue1,2

1Center for Cognition and Brain Disorders, Institute of Psychological Sciences and the Affiliated Hospital, Hangzhou Normal University, Hangzhou 311121, Peoples Republic of China; 2Zhejiang Key Laboratory for Research in Assessment of Cognitive Impairments, Hangzhou 311121, Peoples Republic of China

*These authors contributed equally to this work

Correspondence: Shao-Wei XueCenter for Cognition and Brain Disorders, Hangzhou Normal University, No. 2318, Yuhangtang Road, Hangzhou, Zhejiang 311121, Peoples Republic of ChinaTel/Fax +86-571-28867717Email xuedrm@126.com

Purpose: In recent years, machine learning techniques have received increasing attention as a promising approach to differentiating patients from healthy subjects. Therefore, some resting-state functional magnetic resonance neuroimaging (R-fMRI) studies have used interregional functional connections as discriminative features. The aim of this study was to investigate ADHD-related spatially distributed discriminative features derived from whole-brain resting-state functional connectivity patterns using machine learning.Patients and Methods: We measured the interregional functional connections of the R-fMRI data from 40 ADHD patients and 28 matched typically developing controls. Machine learning was used to discriminate ADHD patients from controls. Classification performance was assessed by permutation tests.Results: The results from the model with the highest classification accuracy showed that 85.3% of participants were correctly identified using leave-one-out cross-validation (LOOV) with support vector machine (SVM). The majority of the most discriminative functional connections were located within or between the cerebellum, default mode network (DMN) and frontoparietal regions. Approximately half of the most discriminative connections were associated with the cerebellum. The cerebellum, right superior orbitofrontal cortex, left olfactory cortex, left gyrus rectus, right superior temporal pole, right calcarine gyrus and bilateral inferior occipital cortex showed the highest discriminative power in classification. Regarding the brainbehaviour relationships, some functional connections between the cerebellum and DMN regions were significantly correlated with behavioural symptoms in ADHD (P < 0.05).Conclusion: This study indicated that whole-brain resting-state functional connections might provide potential neuroimaging-based information for clinically assisting the diagnosis of ADHD.

Keywords: attention deficit hyperactivity disorder, ADHD, resting-state fMRI, R-fMRI, machine learning approach, support vector machine, SVM, leave-one-out cross-validation

This work is published and licensed by Dove Medical Press Limited. The full terms of this license are available at https://www.dovepress.com/terms.php and incorporate the Creative Commons Attribution - Non Commercial (unported, v3.0) License.By accessing the work you hereby accept the Terms. Non-commercial uses of the work are permitted without any further permission from Dove Medical Press Limited, provided the work is properly attributed. For permission for commercial use of this work, please see paragraphs 4.2 and 5 of our Terms.

Here is the original post:
Differentiating Boys with ADHD from Those with Typical Development Bas | NDT - Dove Medical Press

Achieving production-level governance with machine-learning projects currently presents unique challenges. A new space of tools and practices is emerging under the name MLOps. The space is analogous to DevOps but tailored to the practices and workflows of machine learning.

Machine learning models make predictions for new data based on the data they have been trained on. Managing this data in a way that can be safely used in live environments is challenging, and one of the key reasons why 80 per cent of data science projects never make it to production an estimate from Gartner.

It is essential that the data is clean, correct, and safe to use without any privacy or bias issues. Real-world data can also continuously change, so inputs and predictions have to be monitored for any shifts that may be problematic for the model. These are complex challenges that are distinct from those found in traditional DevOps.

DevOps practices are centred on the build and release process and continuous integration. Traditional development builds are packages of executable artifacts compiled from source code. Non-code supporting data in these builds tends to be limited to relatively small static config files. In essence, traditional DevOps is geared to building programs consisting of sets of explicitly defined rules that give specific outputs in response to specific inputs.

In contrast, machine-learning models make predictions by indirectly capturing patterns from data, not by formulating all the rules. A characteristic machine-learning problem involves making new predictions based on known data, such as predicting the price of a house using known house prices and details such as the number of bedrooms, square footage, and location. Machine-learning builds run a pipeline that extracts patterns from data and creates a weighted machine-learning model artifact. This makes these builds far more complex and the whole data science workflow more experimental. As a result, a key part of the MLOps challenge is supporting multi-step machine learning model builds that involve large data volumes and varying parameters.

To run projects safely in live environments, we need to be able to monitor for problem situations and see how to fix things when they go wrong. There are pretty standard DevOps practices for how to record code builds in order to go back to old versions. But MLOps does not yet have standardisation on how to record and go back to the data that was used to train a version of a model.

There are also special MLOps challenges to face in the live environment. There are largely agreed DevOps approaches for monitoring for error codes or an increase in latency. But its a different challenge to monitor for bad predictions. You may not have any direct way of knowing whether a prediction is good, and may have to instead monitor indirect signals such as customer behaviour (conversions, rate of customers leaving the site, any feedback submitted). It can also be hard to know in advance how well your training data represents your live data. For example, it might match well at a general level but there could be specific kinds of exceptions. This risk can be mitigated with careful monitoring and cautious management of the rollout of new versions.

The effort involved in solving MLOps challenges can be reduced by leveraging a platform and applying it to the particular case. Many organisations face a choice of whether to use an off-the-shelf machine-learning platform or try to put an in-house platform together themselves by assembling open-source components.

Some machine-learning platforms are part of a cloud providers offering, such as AWS SageMaker or AzureML. This may or may not appeal, depending on the cloud strategy of the organisation. Other platforms are not cloud-specific and instead offer self-install or a custom hosted solution (eg, Databricks MLflow).

Instead of choosing a platform, organisations can instead choose to assemble their own. This may be a preferred route when requirements are too niche to fit a current platform, such as needing integrations to other in-house systems or if data has to be stored in a particular location or format. Choosing to assemble an in-house platform requires learning to navigate the ML tool landscape. This landscape is complex with different tools specialising in different niches and in some cases there are competing tools approaching similar problems in different ways (see the Linux Foundations LF AI project for a visualization or categorised lists from the Institute for Ethical AI).

The Linux Foundations diagram of MLOps tools ... Click for full detail

For organisations using Kubernetes, the kubeflow project presents an interesting option as it aims to curate a set of open-source tools and make them work well together on kubernetes. The project is led by Google, and top contributors (as listed by IBM) include IBM, Cisco, Caicloud, Amazon, and Microsoft, as well as ML tooling provider Seldon, Chinese tech giant NetEase, Japanese tech conglomerate NTT, and hardware giant Intel.

Challenges around reproducibility and monitoring of machine learning systems are governance problems. They need to be addressed in order to be confident that a production system can be maintained and that any challenges from auditors or customers can be answered. For many projects these are not the only challenges as customers might reasonably expect to be able to ask why a prediction concerning them was made. In some cases this may also be a legal requirement as the European Unions General Data Protection Regulation states that a "data subject" has a right to "meaningful information about the logic involved" in any automated decision that relates to them.

Explainability is a data science problem in itself. Modelling techniques can be divided into black-box and white-box, depending on whether the method can naturally be inspected to provide insight into the reasons for particular predictions. With black-box models, such as proprietary neural networks, the options for interpreting results are more restricted and more difficult to use than the options for interpreting a white-box linear model. In highly regulated industries, it can be impossible for AI projects to move forward without supporting explainability. For example, medical diagnosis systems may need to be highly interpretable so that they can be investigated when things go wrong or so that the model can aid a human doctor. This can mean that projects are restricted to working with models that admit of acceptable interpretability. Making black-box models more interpretable is a fast-growth area, with new techniques rapidly becoming available.

The MLOps scene is evolving as machine-learning becomes more widely adopted, and we learn more about what counts as best practice for different use cases. Different organisations have different machine learning use cases and therefore differing needs. As the field evolves well likely see greater standardisation, and even the more challenging use cases will become better supported.

Ryan Dawson is a core member of the Seldon open-source team, providing tooling for machine-learning deployments to Kubernetes. He has spent 10 years working in the Java development scene in London across a variety of industries.

Bringing DevOps principles to machine learning throws up some unique challenges, not least very different workflows and artifacts. Ryan will dive into this topic in May at Continuous Lifecycle London 2020 a conference organized by The Register's mothership, Situation Publishing.

You can find out more, and book tickets, right here.

Sponsored: Quit your addiction to storage

More:
What would machine learning look like if you mixed in DevOps? Wonder no more, we lift the lid on MLOps - The Register

The work of MIT computer scientist Aleksander Madry is fueled by one core mission: doing machine learning the right way.

Madrys research centers largely on making machine learning a type of artificial intelligence more accurate, efficient, and robust against errors. In his classroom and beyond, he also worries about questions of ethical computing, as we approach an age where artificial intelligence will have great impact on many sectors of society.

I want society to truly embrace machine learning, says Madry, a recently tenured professor in the Department of Electrical Engineering and Computer Science. To do that, we need to figure out how to train models that people can use safely, reliably, and in a way that they understand.

Interestingly, his work with machine learning dates back only a couple of years, to shortly after he joined MIT in 2015. In that time, his research group has published several critical papers demonstrating that certain models can be easily tricked to produce inaccurate results and showing how to make them more robust.

In the end, he aims to make each models decisions more interpretable by humans, so researchers can peer inside to see where things went awry. At the same time, he wants to enable nonexperts to deploy the improved models in the real world for, say, helping diagnose disease or control driverless cars.

Its not just about trying to crack open the machine-learning black box. I want to open it up, see how it works, and pack it back up, so people can use it without needing to understand whats going on inside, he says.

For the love of algorithms

Madry was born in Wroclaw, Poland, where he attended the University of Wroclaw as an undergraduate in the mid-2000s. While he harbored interest in computer science and physics, I actually never thought Id become a scientist, he says.

An avid video gamer, Madry initially enrolled in the computer science program with intentions of programming his own games. But in joining friends in a few classes in theoretical computer science and, in particular, theory of algorithms, he fell in love with the material. Algorithm theory aims to find efficient optimization procedures for solving computational problems, which requires tackling difficult mathematical questions. I realized I enjoy thinking deeply about something and trying to figure it out, says Madry, who wound up double-majoring in physics and computer science.

When it came to delving deeper into algorithms in graduate school, he went to his first choice: MIT. Here, he worked under both Michel X. Goemans, who was a major figure in applied math and algorithm optimization, and Jonathan A. Kelner, who had just arrived to MIT as a junior faculty working in that field. For his PhD dissertation, Madry developed algorithms that solved a number of longstanding problems in graph algorithms, earning the 2011 George M. Sprowls Doctoral Dissertation Award for the best MIT doctoral thesis in computer science.

After his PhD, Madry spent a year as a postdoc at Microsoft Research New England, before teaching for three years at the Swiss Federal Institute of Technology Lausanne which Madry calls the Swiss version of MIT. But his alma mater kept calling him back: MIT has the thrilling energy I was missing. Its in my DNA.

Getting adversarial

Shortly after joining MIT, Madry found himself swept up in a novel science: machine learning. In particular, he focused on understanding the re-emerging paradigm of deep learning. Thats an artificial-intelligence application that uses multiple computing layers to extract high-level features from raw input such as using pixel-level data to classify images. MITs campus was, at the time, buzzing with new innovations in the domain.

But that begged the question: Was machine learning all hype or solid science? It seemed to work, but no one actually understood how and why, Madry says.

Answering that question set his group on a long journey, running experiment after experiment on deep-learning models to understand the underlying principles. A major milestone in this journey was an influential paper they published in 2018, developing a methodology for making machine-learning models more resistant to adversarial examples. Adversarial examples are slight perturbations to input data that are imperceptible to humans such as changing the color of one pixel in an image but cause a model to make inaccurate predictions. They illuminate a major shortcoming of existing machine-learning tools.

Continuing this line of work, Madrys group showed that the existence of these mysterious adversarial examples may contribute to how machine-learning models make decisions. In particular, models designed to differentiate images of, say, cats and dogs, make decisions based on features that do not align with how humans make classifications. Simply changing these features can make the model consistently misclassify cats as dogs, without changing anything in the image thats really meaningful to humans.

Results indicated some models which may be used to, say, identify abnormalities in medical images or help autonomous cars identify objects in the road arent exactly up to snuff. People often think these models are superhuman, but they didnt actually solve the classification problem we intend them to solve, Madry says. And their complete vulnerability to adversarial examples was a manifestation of that fact. That was an eye-opening finding.

Thats why Madry seeks to make machine-learning models more interpretable to humans. New models hes developed show how much certain pixels in images the system is trained on can influence the systems predictions. Researchers can then tweak the models to focus on pixels clusters more closely correlated with identifiable features such as detecting an animals snout, ears, and tail. In the end, that will help make the models more humanlike or superhumanlike in their decisions. To further this work, Madry and his colleagues recently founded the MIT Center for Deployable Machine Learning, a collaborative research effort working toward building machine-learning tools ready for real-world deployment.

We want machine learning not just as a toy, but as something you can use in, say, an autonomous car, or health care. Right now, we dont understand enough to have sufficient confidence in it for those critical applications, Madry says.

Shaping education and policy

Madry views artificial intelligence and decision making (AI+D is one of the three new academic units in the Department of Electrical Engineering and Computer Science) as the interface of computing thats going to have the biggest impact on society.

In that regard, he makes sure to expose his students to the human aspect of computing. In part, that means considering consequences of what theyre building. Often, he says, students will be overly ambitious in creating new technologies, but they havent thought through potential ramifications on individuals and society. Building something cool isnt a good enough reason to build something, Madry says. Its about thinking about not if we can build something, but if we should build something.

Madry has also been engaging in conversations about laws and policies to help regulate machine learning. A point of these discussions, he says, is to better understand the costs and benefits of unleashing machine-learning technologies on society.

Sometimes we overestimate the power of machine learning, thinking it will be our salvation. Sometimes we underestimate the cost it may have on society, Madry says. To do machine learning right, theres still a lot still left to figure out.

See the original post:
Doing machine learning the right way - MIT News

The report on the Machine Learning Software Market is a compilation of intelligent, broad research studies that will help players and stakeholders to make informed business decisions in future. It offers specific and reliable recommendations for players to better tackle challenges in the Machine Learning Software market. Furthermore, it comes out as a powerful resource providing up to date and verified information and data on various aspects of the Machine Learning Software market. Readers will be able to gain deeper understanding of the competitive landscape and its future scenarios, crucial dynamics, and leading segments of the Machine Learning Software market. Buyers of the report will have access to accurate PESTLE, SWOT, and other types of analysis on the Machine Learning Software market.

The Global Machine Learning Software Market is growing at a faster pace with substantial growth rates over the last few years and is estimated that the market will grow significantly in the forecasted period i.e. 2019 to 2026.

Machine Learning Software Market: A Competitive Perspective

Competition is a major subject in any market research analysis. With the help of the competitive analysis provided in the report, players can easily study key strategies adopted by leading players of the Machine Learning Software market. They will also be able to plan counterstrategies to gain a competitive advantage in the Machine Learning Software market. Major as well as emerging players of the Machine Learning Software market are closely studied taking into consideration their market share, production, revenue, sales growth, gross margin, product portfolio, and other significant factors. This will help players to become familiar with the moves of their toughest competitors in the Machine Learning Software market.

Machine Learning Software Market: Drivers and Limitations

The report section explains the various drivers and controls that have shaped the global market. The detailed analysis of many market drivers enables readers to get a clear overview of the market, including the market environment, government policy, product innovation, development and market risks.

The research report also identifies the creative opportunities, challenges, and challenges of the Machine Learning Software market. The framework of the information will help the reader identify and plan strategies for the potential. Our obstacles, challenges and market challenges also help readers understand how the company can prevent this.

Machine Learning Software Market: Segment Analysis

The segmental analysis section of the report includes a thorough research study on key type and application segments of the Machine Learning Software market. All of the segments considered for the study are analyzed in quite some detail on the basis of market share, growth rate, recent developments, technology, and other critical factors. The segmental analysis provided in the report will help players to identify high-growth segments of the Machine Learning Software market and clearly understand their growth journey.

Ask for Discount @ https://www.marketresearchintellect.com/ask-for-discount/?rid=173628&utm_source=NT&utm_medium=888

Machine Learning Software Market: Regional Analysis

This section of the report contains detailed information on the market in different regions. Each region offers a different market size because each state has different government policies and other factors. The regions included in the report are North America, Europe, Asia Pacific, the Middle East and Africa. Information about the different regions helps the reader to better understand the global market.

Table of Content

1 Introduction of Machine Learning Software Market

1.1 Overview of the Market1.2 Scope of Report1.3 Assumptions

2 Executive Summary

3 Research Methodology of Market Research Intellect

3.1 Data Mining3.2 Validation3.3 Primary Interviews3.4 List of Data Sources

4 Machine Learning Software Market Outlook

4.1 Overview4.2 Market Dynamics4.2.1 Drivers4.2.2 Restraints4.2.3 Opportunities4.3 Porters Five Force Model4.4 Value Chain Analysis

5 Machine Learning Software Market , By Deployment Model

5.1 Overview

6 Machine Learning Software Market , By Solution

6.1 Overview

7 Machine Learning Software Market , By Vertical

7.1 Overview

8 Machine Learning Software Market , By Geography

8.1 Overview8.2 North America8.2.1 U.S.8.2.2 Canada8.2.3 Mexico8.3 Europe8.3.1 Germany8.3.2 U.K.8.3.3 France8.3.4 Rest of Europe8.4 Asia Pacific8.4.1 China8.4.2 Japan8.4.3 India8.4.4 Rest of Asia Pacific8.5 Rest of the World8.5.1 Latin America8.5.2 Middle East

9 Machine Learning Software Market Competitive Landscape

9.1 Overview9.2 Company Market Ranking9.3 Key Development Strategies

10 Company Profiles

10.1.1 Overview10.1.2 Financial Performance10.1.3 Product Outlook10.1.4 Key Developments

11 Appendix

11.1 Related Research

Request Report Customization @ https://www.marketresearchintellect.com/product/global-machine-learning-software-market-size-forecast/?utm_source=NT&utm_medium=888

About Us:

Market Research Intellect provides syndicated and customized research reports to clients from various industries and organizations with the aim of delivering functional expertise. We provide reports for all industries including Energy, Technology, Manufacturing and Construction, Chemicals and Materials, Food and Beverage and more. These reports deliver an in-depth study of the market with industry analysis, market value for regions and countries and trends that are pertinent to the industry.

Contact Us:

Mr. Steven FernandesMarket Research IntellectNew Jersey ( USA )Tel: +1-650-781-4080

Email: [emailprotected]

TAGS: Machine Learning Software Market Size, Machine Learning Software Market Growth, Machine Learning Software Market Forecast, Machine Learning Software Market Analysis, Machine Learning Software Market Trends, Machine Learning Software Market

Read the original here:
Machine Learning Software Market Increasing Demand with Leading Player, Comprehensive Analysis, Forecast to 2026 - News Times

By not touching your face, you reduce the chances of getting sick from a virus or bacteria. This website, called Do Not Touch Your Face, uses your webcam to analyze your face and alert you with a tone if it catches you touching your face.

From the FAQ

How does this work?

Using your webcam, you train a machine learning algorithm (specificallyTensorflow.js) to recognize you touching your face and not touching your face. Once it's trained, it watches and alerts you when you touch your face.

Why shouldn't I touch my face?

TheCDC recommendsnot touching your face as one action you can take to prevent getting COVID-19. Other things you should do: stay home if you're sick and avoid contact with other sick people. But you probably knew that already.

The alerts aren't working!

Try refreshing the page and trying again. Every time you reload the page, the algorithm retrains itself.

Do you keep my information?

Nope. This entire site runs locallyall the calculations from your webcam and alerts are done on your computer and are never sent over the internet.

Will this stop me from getting COVID-19?

Not for sure, but it might help.

Who made this?

This was made with love and fear byMike Bodge,Brian Moore, andIsaac Blankensmith. Be safe out there.

Cleethorpes is a faded northern English resort town whose inherent grimness is leavened by low rainfall and a nice sandy beach. And now it is to become home to a giant white metal palm tree, to the dismay of some locals. Artist Wolfgang Weileder has said the sculpture will serve as a warning for the []

Since the 60s Genesis P-Orridge has been one of the masterminds behind artist collective COUM Transmissions and seminal music acts Throbbing Gristle and Psychic TV. Beyond that, P-Orridge has had an astonishing career in the visual arts, founding an artist collective called Thee Temple ov Psychick Youth, as well as helming the infamous pandrogeny project []

Colossal writes: Designed to recycle outdated electronics, multiple musical projects by Electronicos Fantasticos utilize a version of the barcode system found on every package on store shelves. When scanned, each pattern sends a signal to its audio component, emitting the corresponding sound wave. The black and white stripes produce a variety of rhythmic and tonal []

In an age where blockbuster MMOs and aggressive action-adventure games dominate the landscape, theres always something to be said for smart, atmospheric, slow-burn gaming that truly forces players to stretch their minds rather than their firepower to notch a victory. Thats why the sci-fi themed, first-person puzzler Lightmatter has already started building a following as []

Tech moves so fast that practically the minute you lift the latest, fastest, most tricked-out new laptop on the market off the store shelf, the staff is filling that space with a newer, faster, even more, tricked-out model. Thats just the speed of advancement and that march is unstoppable. So instead of paying a []

Back in 2007, high schooler Mike Radenbaugh got tired of pumping his old bicycle back and forth to campus every day. Instead, he pulled together some parts, attached an electric motor to his bike and his first e-bike was born. It wouldnt be his last. 13 years later, Radenbaugh heads up Seattle-based Rad Power []

More:
This website uses machine learning and your webcam to train you not to touch your face - Boing Boing

Campaigns & Elections Reed Award winners represent the best-of-the-best in the political campaign and advocacy industries. CampaignTesters' proprietary platform aims to deliver key audience insights for organizations to validate, revise and perfect their video content messaging.

CampaignTester is a cutting-edge mobile-based platform that utilizes emotion analytics and machine learning to detect a users emotion and engagement level while watching video content. Their proprietary platform aims to deliver key audience insights for organizations to validate, revise and perfect their video content messaging.

Campaigns & Elections Reed Award winners represent the best-of-the-best in the political campaign and advocacy industries. The 2020 Reed Awards honored winners across 16 distinct category groups, representing the different specialisms of the political campaign industry, with distinct category groups for International (non-US) work, and Grassroots Advocacy work.

It was particularly meaningful being recognized among some of the finest marketers and technologists in the world, Bill Lickson, CampaignTesters Chief Operating Officer affirmed. I was thrilled and honored to accept this prestigious award on behalf of our entire talented team.

Aaron Itzkowitz, Chief Executive Officer and Founder of CampaignTester added, This award is a great start to what looks to be a wonderful year for our client-partners and our company. While our technology was recognized for excellence in political marketing, our technology is for any industry that uses video in marketing.

About Campaigns & Elections Reed AwardsThe Campaigns & Elections Reed Awards, named after Campaigns & Elections founder Stanley Foster Reed, recognizes excellence in political campaigning, campaign management, political consulting and political design, grassroots & advocacy.

For more information about CampaignTester, visit CampaignTester.com or contact Press@campaigntester.com, 352-247-7865

Read the original here:
Innovative AI and Machine-Learning Technology That Detects Emotion Wins Top Award - PR.com

Were now in the midst of theFourth Industrial Revolution,where humans work side-by-side with machines. This means business users must attain new digital skills to effectively supervise the huge influx of digital workers, driven by the rise of robotic process automation (RPA) software robots.

Its even more significant given that companies deploying these bots expect an increase by as much as 50 percent over the next two years, according to IDC in itsContent Intelligence: For the Future of Worksurvey.

The reason were seeing more software robots within the enterprise is that transformation initiatives are no longer solely owned by the IT department. Instead we are seeing organizations form a new Center of Excellence (COE) where multiple people within the organization are involved in the automation process, capturing C-Level visibility and engagement.These COEs are growing at a pace, with business analysts becoming central to the process of assessment and use of RPA tools to facilitate change.

Historically, AI technologies like machine learning have been difficult to incorporate, but now next generation applications are packaging AI technology in a way that is easy to train and consume in order to build and extend the digital workforce. But while teaching the new software robot no longer requires a developer in AI or machine learning, it does mean that business analysts will need to gain more skills to be proficient in process assessment methods, designing, training, deploying, and managing the new digital workforce.

A concerning 75 percent of global enterprises in IDCs Future of Work report said it was difficult to recruit people with new digital skills needed for transformation, and 20 percent cited inadequate worker training was a leading challenge. As we enter a new decade, business analysts with these higher skills are therefore crucial more than ever to adequately supervise and train digital workers.

So how can we prepare our workforce for the new digital era? There are two approaches: advocacy and access.

Lets be clear, the digital skills gap Im speaking of is not the transition we saw at the turn of the century where people traded their filing cabinets and typewriters for personal computers and Microsoft Word. Over the last decade enterprises have successfully transitioned to business process automation solutions in virtually every department from shipping, legal, accounts payable, payroll, human resources and recruiting, to sales and marketing and customer service. Todays workforce is proficient in using software for automation and collaboration.

Despite the advances in automation however, employees are often still performing manual work that falls outside of these systems especially when these processes involve unstructured content documents, images, text, and emails.

Now, with advances in AI, robots can be trained to carry out this manual work through the use of specific pre-packaged advanced skills. Also, by showing the bots how to perform a task, pointing out where they went wrong so they learn from their mistakes, they effectively gain human understanding such as thinking and reasoning so they become subject matter experts.

This is a key stage of the automation process and why it is important that business analysts understand how AI for content can be applied and incorporated into their procedures.

They will need access and knowledge to tools that can understand a business process and make recommendations. They can use these tools that apply AI to processing content but do it in such a way that does not require an advanced degree around machine learning and other AI technologies. This type of training and digital knowledge is imperative as companies move forward with automation. By doing this, it frees up more time for the employees to concentrate on more complex tasks or important business activities, like improving customer service. It is enabling and empowering more people in the organization, not just a few people with tribal knowledge who know the systems.

Take the role of compliance officers, for example. The challenge for these employees is sifting through the amount of documents and data associated with Know Your Customer (KYC) and Anti-Money Laundering (AML) checks as part of customer due diligence process requirements. Without performing thorough checks, banks are at serious risks at incurring hefty fines in the millions.

Many banks are using RPA as a first step to automate the collecting of documents and data, but still leave it up to the compliance officer to sift through documents and find the data that is relevant to their decisions.

By allowing robots to read the contracts and pick out relevant data, the compliance officer and the legal team can focus on the higher value work rather than manually inputting data into software or searching for key phrases. Digital skills will help a wide variety of professionals to augment and improve their work productivity from the legal team, HR, accounts payable, claims adjustors and more.

Equipping your workforce with new skills that complement the digital world of business today has mainly been industry driven. Some of the major RPA vendors such as Blue Prism and UiPath offer conferences, seminars and webinars that teach you how to train digital workers with Content Intelligence skills. Likewise, AI-enabling companies also offer the same resources for working with cognitive skills no matter which RPA platform you choose. Other resources are value-added resellers and integrated solution partners who will work with your automation team to deploy digital workers and train them to maximize their efficiency.

As organizations become more comfortable and proficient with training bots with specific cognitive skills, youll soon find internal marketplaces emerge within companies where departments and business units across the entire organization can share and access them.

Universities are also catching up to the speed of business and offering courses to equip the next generation of business and management graduates. There are notably two universities offering software robotics courses.California State University at Fullertons Mihaylo College of Business and Economicsis offering both graduate and undergraduate courses explaining the applications of RPA to drive efficiencies and improve performance in accounting. As part of a partnership with UiPath, the course features presentations and applied demonstrations from experts in the field including professionals from the Big Four accounting firms.

From the technology perspective,Carnegie Mellon Universitys Heinz CollegeSchools of Information Systems & Managementis offering an Advances in Robotic Process Information coursefor its Masters Program in the spring of 2020. Technology leaders will share their experiences and givestudentsaccess to the latest artificial intelligence and machine learning technology, such as RPA tools from Blue Prism and Content Intelligence skills from ABBYY.

With a new set of digital skills business analysts will heighten the level of automation in their company so staff can focus on more higher-value tasks that require emotional intelligence qualities such as judgement, discernment and empathy. Its a sophisticated balance between being efficient in their core profession, understanding the organizations needs and being digitally proficient to embrace the future of work. In turn,businessescan expect increased employee productivity and be on their way to have a better pulse on their overall Digital Intelligence having a complete understanding of how their business operates and to allocate resources and improve operations based on facts.

Related: Smart Tech is Changing Apartment Living

The piece Are Business Analysts Ready for the New Digital Era? by Bill Galusha first appeared on Innovation & Tech Today.

More here:
Are Business Analysts Ready for the New Digital Era? - Grit Daily

Abstract

The unusual correlated state that emerges in URu2Si2 below THO = 17.5 K is known as hidden order because even basic characteristics of the order parameter, such as its dimensionality (whether it has one component or two), are hidden. We use resonant ultrasound spectroscopy to measure the symmetry-resolved elastic anomalies across THO. We observe no anomalies in the shear elastic moduli, providing strong thermodynamic evidence for a one-component order parameter. We develop a machine learning framework that reaches this conclusion directly from the raw data, even in a crystal that is too small for traditional resonant ultrasound. Our result rules out a broad class of theories of hidden order based on two-component order parameters, and constrains the nature of the fluctuations from which unconventional superconductivity emerges at lower temperature. Our machine learning framework is a powerful new tool for classifying the ubiquitous competing orders in correlated electron systems.

Phase transitions mark the boundary between different states of matter, such as liquid to solid, or paramagnet to ferromagnet. At the phase transition, the system lowers its symmetry: Translationally invariant liquids become crystalline solids; paramagnetic spins align to break time reversal and rotation symmetry in a magnet. The conventional description of a second-order phase transitionLandau theoryrequires knowledge of which symmetries are broken in the low-temperature phase to construct an order parameter (OP). Several possibilities have been put forth for the symmetry of the OP in the hidden order (HO) state of URu2Si2 (Table 1), but most of these rely on specific microscopic mechanisms that are difficult to verify experimentally (1, 2).

Note that designations such as hexadecapole order are only applicable in free spacecrystalline electric fields break these large multipoles into the representations listed in this table.

The purpose of this paper was to use resonant ultrasound spectroscopy (RUS) to place strict thermodynamic constraintsindependent of microscopic mechanismon the OP symmetry in URu2Si2. While RUS is a powerful techniquecapable of constraining or identifying the symmetries broken at a phase transition (3)it has one substantial drawback: A single missing resonance renders an entire spectrum unusable. This is because traditional RUS data analysis relies on solving the elastic wave equation and mapping the computed resonances one to one with measured resonancesa single missing resonance invalidates this mapping. Here, we develop a new machine learningbased approach. We take advantage of the fact that neural networks can be trained to recognize features in complex datasets and classify the state of matter that produces such data (49). We validate this approach by analyzing an RUS dataset that we are confident can also be analyzed using traditional methods (data from a large single-crystal URu2Si2 with a well-defined geometry). We then analyze data from a higher-quality URu2Si2 sample that has an ill-defined geometrya task that is impossible for the traditional analysis method but which is easily performed by our neural network.

While the broken symmetries of HO are unknown, most theories assume some form of multipolar order, whereby localized 5f electrons on the uranium site occupy orbitals that order below THO = 17.5 K. However, direct experimental evidence for localized 5f electronssuch as crystalline electric field level splittingdoes not exist (1), leaving room for theories of HO based on itinerant 5f electrons. Many possible OPs remain in contention, but, whether itinerant or localized, all theories of HO can be classified on the basis of the dimensionality of their point group representation: one component (1019) or two component (2026) [see Table 1 and (27)]. Theories of two-component OPs are motivated largely by the experiments of Okazaki et al. (28) and Tonegawa et al. (29), which detect a small C4 symmetry breaking at THO. More recent x-ray experiments have cast doubt on these results (30), leaving even the dimensionality of the OP in URu2Si2 an open question.

Determining OP dimensionality is more than an exercise in accounting: The two-component nature of loop currents allows for dynamics that have been suggested to explain the pseudogap in high-Tc cuprates (31), and the proposed two-component px + ipy superconducting state of Sr2RuO4 has a unique topological structure that can support Majorana fermions (32, 33). Establishing the dimensionality of the HO state not only allows us to rigorously exclude a large number of possible OPs but also provides a starting point for understanding the unusual superconductivity that emerges at lower temperature in URu2Si2.

RUS measures the mechanical resonance frequencies of a single-crystal specimenanalogous to the harmonics of a guitar string but in three dimensions (see Fig. 1A). A subset of this spectrum for a 3 mm by 2.8 mm by 2.6 mm crystal of URu2Si2 (sample S1) is shown in Fig. 1B, with each peak occurring at a unique eigenfrequency of the elastic wave equation (see the Supplementary Materials). Encoded within these resonances is information about the samples dimensions and density, which are known, and the six elastic moduli, which are unknown. As electrons and phonons are coupled strongly in metals, the temperature dependence of the elastic moduli reveals fluctuations and instabilities in the electronic subsystem. In particular, elastic moduli are sensitive to symmetry breaking at electronic phase transitions (3, 34). The difficulty lies in converting the temperature dependence of the resonance spectrum into the temperature dependence of the elastic moduli. The traditional analysis involves solving the three-dimensional (3D) elastic wave equation and adjusting the elastic moduli to match the experimental resonance spectrum. However, if even a single resonance is missing from the spectrum (e.g., due to weak coupling of a particular mode to the transducers), then this analysis scheme breaks down [see Ramshaw et al. (3) for further discussion of this problem].

(A) Schematic resonance eigenmodes obtained as a solution to the 3D elastic wave equation. Each mode contains a unique proportion of the five irreducible strains (see Fig. 2A). (B) Room temperature ultrasonic spectrum of sample S1, shown between 500 kHz and 1 MHz. (C) Temperature evolution of seven characteristic resonances, out of 29 total measured resonances, near the HO transitionplots are shifted vertically for clarity. Three resonances (672, 713, and 1564 kHz) show jumps at THO (inset illustrates what is meant by the jump), while the others do not, signifying contributions from different symmetry channels.

Figure 1C shows the temperature dependence of seven representative elastic resonances through THO (29 resonances were measured in total). Note that while some resonances show a step-like discontinuity or jump at THO, others do not. This jump is present in the elastic moduli for all second-order phase transitions (3, 34, 35) but has never before been observed in URu2Si2 due to insufficient experimental resolution (3640). Traditional RUS produces spectra at each temperature, such as the one shown in Fig. 1B, by sweeping the entire frequency range using a lock-in amplifier. The resonance frequencies are then extracted by fitting Lorentzians to each peak (34). We have developed a new approach whereby the entire spectrum is swept only onceto identify the resonancesand then, each resonance is tracked as a function of temperature with high precision using a phase-locked loop. This increases the density of data points per unit temperature by roughly a factor of 1000 and increases the signal-to-noise by a factor of 30 (see Materials and Methods).

The complex strain fields produced at each resonance frequency (Fig. 1A) can be broken down locally into irreducible representations of strain (k). Each irreducible strain then couples to an OP of a particular symmetry in a straightforward manner (35). In this way, analysis of the temperature dependence of the resonance frequencies can identify or constrain the OP symmetry. In a tetragonal crystal, such as URu2Si2, elastic strain breaks into five irreducible representations (Fig. 2): two compressive strains transforming as the identity A1g representation, and three shear strains transforming as the B1g, B2g, and Eg representations. Allowed terms in the free energy are products of strains and OPs that transform as the A1g representation. As HO is thought to break at least translational symmetry, the lowest-order terms allowed by both one-component and two-component OPs are linear in the A1g strains and quadratic in OP: = A1g 2 [see (41)]. Quadratic-in-order-parameter, linear-in-strain coupling produces a discontinuity in the associated elastic modulus at the phase transition: This jump is related to discontinuities in the specific heat and other thermodynamic quantities through Ehrenfest relations (34, 42). For OPs with one-component representations (any of the Ai or Bi representations of D4h), only the elastic moduli corresponding to A1g compressional strains couple in this manner. In contrast, shear strains couple as F=k22 and show at most a change in slope at THO (3). Thus c33, c23, and (c11 + c12)/2 may exhibit jumps at phase transitions corresponding to one-component OPs, while (c11 c12)/2, c66, and c44 cannot.

(A) The tetragonal crystal structure of URu2Si2 and its five irreducible representations of strain, along with the associated moduli. Each resonance shown in Fig. 1A can be decomposed into this basis set of strains, modulated in phase at long wavelengths throughout the crystal. c23 characterizes the direct coupling between the two A1g strains. (B) Compressional (A1g shown in orange) and (C) shear (B1g, B2g, and Eg shown in blue) elastic moduli, with dashed guides to the eye showing the temperature dependence extrapolated from below and above THO. The absolute values (in gigapascals) of the moduli at 20 K were determined to be (c11 + c12)/2 = 218.0, c33 = 307.4, c23 = 112.8, (c11 c12)/2 = 65.2, c66 = 140.6, and c44 = 101.8. (D) The magnitude of the jumps at THO with their experimental uncertainties. A large jump occurs in (c11 + c12)/2 at THO, along with a small jump in c23. The shear moduli, on the other hand, show only a change in slope at THOthis constrains the OP of the HO state to transform as a one-component representation.

Two-component OPs (of the Ei representations), on the other hand, have bilinear forms that can couple with two of the shear strains to first order. A two-component OP, ={x,y}, has the bilinears x2+y2,x2y2, and xy of the A1g, B1g, and B2g representations, respectively. In addition to the standard A1gx2+y2 terms, the free energy now contains the terms B1g(x2y2) and B2g xy. A second-order phase transition characterized by a two-component OP therefore exhibits discontinuities in the B1g and B2g shear elastic moduli [(c11 c12)/2 and c66, respectively], in addition to jumps in the compressional A1g moduli (see the Supplementary Materials for a discussion of the E3/2,g representation, pertaining to hastatic order).

We first perform a traditional RUS analysis, extracting the temperature dependence of the six elastic moduli (Fig. 2, B and C) from 29 measured resonances by solving the elastic wave equation and fitting the spectrum using a genetic algorithm [see the Supplementary Materials of Ramshaw et al. (3) for details]. The evolution of the elastic moduli across THO shows jumps in two of the A1g elastic moduli, whereas the B1g and B2g shear moduli show only a break in slope at THO to within our experimental uncertainty (Fig. 2D). Jumps in the shear moduli would be expected for any OP of the two-component Ei representations (2026)the fact that we do not resolve any shear jumps constrains the OP of the HO phase to belong to a one-component representation of D4h. The fact that we do not resolve a jump in c33 is consistent with the magnitudes of the jumps in (c11 + c12)/2 and c23 (see the Supplementary Materials for details).

In principle, this traditional analysis is sufficient to determine the order-parameter dimensionality in URu2Si2. The process of solving for the elastic moduli, however, incorporates systematic errors arising from sample alignment, parallelism, dimensional uncertainty, and thermal contraction. Even more detrimental is the possibility that the measured spectrum is missing a resonance, rendering the entire analysis incorrect. While we are confident in our analysis for the particularly large and well-oriented sample S1, large samples of URu2Si2 are known to be of slightly lower quality (43). Smaller, higher-quality crystals of URu2Si2 do not lend themselves well to RUS studies, being hard to align and polish to high precision. Smaller samples also produce weaker RUS signals, making it easier to miss a resonance. We have therefore developed a new method for extracting symmetry information directly from the resonance spectrum, without needing to first extract the elastic moduli themselves, even if the spectrum is incomplete. This method takes advantage of the power of machine learning algorithms to recognize patterns in complex datasets.

Artificial neural networks (ANNs) are popular machine learning tools due to their ability to classify objects in highly nonlinear ways. In particular, ANNs can approximate smooth functions arbitrarily well (44). Here, we train an ANN to learn a function that maps the jumps in ultrasonic resonances at a phase transition to one of two classes, corresponding to either a one-component or two-component OP. One-component OPs induce jumps only in compressional elastic moduli, whereas two-component OPs also induce jumps in two of the shear moduli. Phase transitions with two-component OPs should therefore show jumps in more ultrasonic resonances at a phase transition than phase transitions with one-component OPs. Our intent is that this difference in the distribution of jumps can be learned by an ANN to discriminate between one-component and two-component OPs.

An ANN must be trained with simulated data that encompass a broad range of possible experimental scenarios. In our case, we simulate RUS spectra given assumptions about the sample and the OP dimensionality. Starting with a set of parameters randomly generated within bounds that we specifythese include the sample geometry, density, and the six elastic moduliwe solve the elastic wave equation to produce the first N resonance frequencies that would be measured in an RUS experiment. Then, using a second set of assumptionswhether the OP has one component or two, whether our simulated experiment has k missing resonances, and the relative sizes of the elastic constant jumps produced at THOwe calculate the jumps at THO for the first n resonances (see Fig. 3). By varying the input assumptions, we produce a large number of training datasets that are intended to encompass the (unknown) experimental parameters.

Values for elastic moduli and dimensions are chosen randomly from a range that bounds our experimental uncertainties. One-component OPs give jumps only in A1g moduli, whereas two-component OPs also give jumps in B1g and B2g moduli. Separate output files are generated corresponding to one-component and two-component OPs, each containing n jumps, where n is the number of frequencies whose temperature evolution could be experimentally measured. We use scaled RUS frequency shifts fj/fj as input to the ANN. The neurons in the hidden layer have weights wij and biases bi. Each output neuron corresponds to one of the two OP dimensionalities under consideration, i.e., one-component and two-component. The output value of each neuron is the networks judgment on the likelihood of that OP dimensionality.

While the sample geometry, density, and moduli are well determined for sample S1 and only varied by a few tens of percent, the dimensionality of the OP, the number of missing resonances, and the sizes of the jumps in each symmetry channel are taken to be completely unknown. We vary these latter parameters across a broad range of physically possible values (see Fig. 3 and the Supplementary Materials for further details). To prepare the simulated data for interpretation by our ANN, we take the first n jumps, sort the jumps by size, normalize the jumps to lie between zero and one, and label the datasets by the dimensionality of the OP that was used to create themeither one component or two.

This normalized and sorted list of numbers {fi/fi} is used as input to an ANN. Our ANN architecture is a fully connected, feedforward neural network with a single hidden layer containing 20 neurons (see Fig. 3). Each neuron j processes the inputs {fi/fi} according to the weight matrix wji and the bias vector bj specific to that neuron as (wjixi + bj), where the rectified linear activation function is given by (y) max (y,0). The sum of the neural outputs is normalized via a softmax layer.

We train the ANN using 10,000 sets of simulated RUS data for the case of a one-component OP, with varied elastic constants, sample geometries, jump magnitudes, and missing resonances, and another 10,000 sets for the case of a two-component OP. We use cross-entropy as the cost function for stochastic gradient descent. We train 10 different neural networks in this way to an accuracy of 90% and then fix each individual networks weights and biases. Once the networks are trained, we ask each ANN for its judgment on the OP dimensionality associated with an experimentally determined set of 29 jumps and average the responses from each neural network. The sizes of the jumps depend on how THO is assignedassigning THO artificially far from the actual phase transition will produce large jumps in all resonances. We therefore repeat our ANN determination using a range of THO around the phase transition and plot the outcome as a function of THO.

Figure 4A shows the results of our ANN analysis for sample S1the same sample discussed above using the traditional analysis. To visually compare the training and experimental data in a transparent fashion, we plot the list of sorted and normalized jumps against their indices in the list. The average of the one-component training data is shown in red; the average of the two-component training data is shown in blue; the experimental jumps are shown in gray. It is clear that the experimental data resemble the one-component training data much more closely. This resemblance is quantified in the inset, showing the ANN confidence that the experimental data belong to the one-component class for varying assignments of THO. We find that the confidence of a one-component OP is maximized in the region of assigned THO that corresponds to the experimental value of THO.

Upper blue curves show the averaged, sorted, simulated frequency shift (jump) data plotted against its index in the sorted list for a two-component OP for (A) sample S1 and (B) sample S2. The data are normalized to range from 0 to 1. Lower, red curves shows the same for a one-component OP. Gray dots show experimental data for critical temperature assignment (A) THO = 17.26 K and (B) THO = 17.505 K, which visually aligns more closely with the average one-component simulated data than the two-component simulations. Insets: Percentage confidence of the one-component output neuron for various assignments of THO averaged over 10 trained networks. A maximum confidence of (A) 83.2% occurs for THO = 17.26 K, and (B) 89.7% for THO = 17.505 K. Sample S2 has a higher value of THO due to its lower impurity concentration, as verified independently by the resistivity. Photo credit: Sayak Ghosh, Cornell University.

Thus far, we have shown that both methodsthe traditional method of extracting the elastic moduli using the elastic wave equation and our new method of examining the resonance spectrum directly using a trained ANNagree that the HO parameter of URu2Si2 is one component. We can now use the neural network to analyze a smaller, irregular-shaped but higher-quality [higher THO (43)] sample that cannot be analyzed using the traditional method due to its complicated geometry. Figure 4B shows the result of the ANN analysis performed on a resonance spectrum of sample S2. The sorted and normalized spectrum looks very similar to that of sample S1, and the averaged ANN outcome gives 90% confidence that the OP is one component. Despite the fact that sample S2 has a geometry such that the elastic moduli cannot be extracted, its resonance spectrum still contains information about the OP dimensionality, and our ANN identifies this successfully.

Our two analyses of ultrasonic resonances across THO in URu2Si2 strongly support one-component OPs, such as electric-hexadecapolar order (14), the chiral density wave observed by Raman spectroscopy (17, 18, 45), and are consistent with the lack of C4 symmetry breaking observed in recent x-ray scattering experiments (30). Our analysis rules against two-component OPs, such as rank-5 superspin (19, 22) and spin nematic order (24). The power of our result lies in its independence from the microscopic origin of the OP: Group theoretical arguments alone are sufficient to rule out large numbers of possible OPs. It could be argued that the coupling constants governing the jumps in the shear moduli are sufficiently small such that the jumps are below our experimental resolution. Previous experiments, however, have shown these coupling constants to be of the same order of magnitude in other materials with multicomponent OPs (35, 46, 47). It has also been demonstrated that the size of the jump in heat capacity at THO is largely insensitive to residual resistivity ratio (RRR) (43, 48, 49). It is therefore hard to imagine that higher RRR samples would yield jumps in the shear moduli.

The use of ANNs to analyze RUS data represents an exciting opportunity to reexamine ultrasound experiments that were previously unable to identify OP symmetry. For example, irregular sample geometry prevented identification of the OP symmetry in the high-Tc superconductor YBa2Cu3O6.98 (34). Reanalysis of this spectrum using our ANN could reveal whether the OP of the pseudogap is associated with Eu-symmetry orbital loop currents. The proposed two-component px + ipy superconducting state of Sr2RuO4 and other potential spin-triplet superconductors could also be identified in this fashion, where traditional pulse-echo ultrasound measurements have been confounded by systematic uncertainty (50).

Beyond RUS, there are many other data analysis problems in experimental physics that stand to be improved using an approach similar to the one presented here (51). In particular, any technique where simulation of a dataset is straightforward but where fitting is difficult should be amenable to a framework of the type used here. The most immediately obvious technique where our algorithm could be applied is nuclear magnetic resonance (NMR) spectroscopy. NMR produces spectra in a similar frequency range to RUS but which originate in the spin-resonances of nuclear magnetic moments. Modern broadband NMR can produce complex temperature-dependent spectra, containing resonances from multiple elements situated at different sites within the unit cell. Given a particular magnetic order, it is relatively straightforward to calculate the NMR spectrumi.e., to produce training data. The inverse problem, however, is more challenging: recovering a temperature-dependent magnetic structure from an NMR dataset. In a way similar to RUS, missing resonances and resonances mistakenly attributed to different elements can render an analysis entirely invalid. It should be relatively straightforward to adapt our framework for generating training data and our ANN to extract temperature (or magnetic field)dependent magnetic structures from NMR spectra.

Sample S1 was grown by the Czochralski method. A single crystal oriented along the crystallographic axes was polished to dimensions 3.0 mm by 2.8 mm by 2.6 mm, with 2.6 mm along the tetragonal long axis. Sample S2 was grown was grown by the Czochralski method and then processed by solid-state electrorefinement. Typical RRR values for ab-plane flakes of URu2Si2 taken from the larger piece range from 100 to 500. The RRR values measured on larger pieces (Fig. 4) are between 10 and 20. For a comparison of different growth methods for URu2Si2 see Gallagher et al. (49).

Resonant ultrasound experiments were performed in a custom-built setup consisting of two compressional-mode lithium niobate transducers, which were vibrationally isolated from the rest of the apparatus. The top transducer was mounted on a freely pivoting arm, ensuring weak coupling and linear response. The response voltage generated on the pickup transducermaximum whenever the drive frequency coincides with a sample resonancewas measured with lock-in technique. The response signal was preamplified using a custom-made charge amplifier to compensate for signal degradation in coaxial cables (52). Oxford Instruments He4 cryostat was used for providing temperature control.

Supplementary material for this article is available at http://advances.sciencemag.org/cgi/content/full/6/10/eaaz4074/DC1

Phase-locked loop

Training data for ANN

Symmetry and coupling

Lack of c33 jump

Resolving the origin of jumps

Compositions of resonances

Resistance measurement

Possible effects from parasitic antiferromagnetism

Table S1. Calculated discontinuities (jumps) in elastic moduli for one- and two-component OPs in a tetragonal system.

Fig. S1. Resonant ultrasound using phase-locked loop.

Fig. S2. Three representative resonance frequencies of URu2Si2 and their attenuation through THO.

Fig. S3. Elastic moduli of URu2Si2 with the contribution above THO subtracted.

Fig. S4. Fitting temperature evolution of resonances.

Fig. S5. Resistance of sample S2 measured from 300 K down to 2 K.

References (5559)

This is an open-access article distributed under the terms of the Creative Commons Attribution-NonCommercial license, which permits use, distribution, and reproduction in any medium, so long as the resultant use is not for commercial advantage and provided the original work is properly cited.

Y. Yamaji, T. Yoshida, A. Fujimori, M. Imada, Hidden self-energies as origin of cuprate superconductivity revealed by machine learning. arXiv:1903.08060 [cond-mat.str-el] (19 March 2019).

It is also generally agreed that OP of the HO state orders at a finite wavevector of Q = (0,0,1/2). Because our measurement occurs close to Q = 0, i.e., at long wavelength, we are only concerned with the point group symmetry of the OP and not with its modulation in space.

There are two A1g strains, xx + yy and zz with associated moduli (c11 + c12)/2 and c33, as well as linear coupling between these two strains that produces a third modulus c23. To simplify the notation, we drop the sum over all three of these terms in the free energy.

B. C. Csji, Approximation with Artificial Neural Networks, thesis, Faculty of Sciences, Etvs Lornd University, Hungary (2001).

S. Altmann, P. Herzig, Point-Group Theory Tables (Clarendon Press, 1994).

Acknowledgments: We thank P. Coleman, P. Chandra, and R. Flint for the helpful discussions. Funding: Work at Los Alamos National Laboratory was performed under the auspices of the U.S. Department of Energy (DOE), Office of Basic Energy Sciences, Division of Materials Sciences and Engineering. M.M., B.J.R., and S.G. acknowledge support by the Cornell Center for Materials Research with funding from the NSF MRSEC program (DMR-1719875). M.M. acknowledges support by the NSF [Platform for the Accelerated Realization, Analysis, and Discovery of Interface Materials (PARADIM)] under cooperative agreement no. DMR-1539918. E.-A.K. acknowledges support from DOE DE-SC0018946. B.J.R. and S.G. acknowledge funding from the NSF under grant no. DMR-1752784. Author contributions: S.G. and B.J.R. designed the experiment. R.B. and E.D.B. grew sample S2. J.A.M. grew sample S1. S.G. acquired and analyzed the data. M.M. and E.-A.K. designed the ANN. K.A.M., A.S., M.M., S.G., and B.J.R. wrote the manuscript with input from all coauthors. Competing interests: The authors declare that they have no competing interests. Data and materials availability: All data needed to evaluate the conclusions in the paper are present in the paper and/or the Supplementary Materials. Additional data related to this paper may be requested from the authors.

See the original post here:
One-component order parameter in URu2Si2 uncovered by resonant ultrasound spectroscopy and machine learning - Science Advances