For all that neural networks can accomplish, we still dont really understand how they operate. Sure, we can program them to learn, but making sense of a machines decision-making process remains much like a fancy puzzle with a dizzying, complex pattern where plenty of integral pieces have yet to be fitted.

If a model was trying to classify an image of said puzzle, for example, it could encounter well-known, but annoying adversarial attacks, or even more run-of-the-mill data or processing issues. But a new, more subtle type of failure recently identified by MIT scientists is another cause for concern: overinterpretation, where algorithms make confident predictions based on details that dont make sense to humans, like random patterns or image borders.

This could be particularly worrisome for high-stakes environments, like split-second decisions for self-driving cars, and medical diagnostics for diseases that need more immediate attention. Autonomous vehicles in particular rely heavily on systems that can accurately understand surroundings and then make quick, safe decisions. The network used specific backgrounds, edges, or particular patterns of the sky to classify traffic lights and street signs irrespective of what else was in the image.

The team found that neural networks trained on popular datasets like CIFAR-10 and ImageNet suffered from overinterpretation. Models trained on CIFAR-10, for example, made confident predictions even when 95 percent of input images were missing, and the remainder is senseless to humans.

Overinterpretation is a dataset problem that's caused by these nonsensical signals in datasets. Not only are these high-confidence images unrecognizable, but they contain less than 10 percent of the original image in unimportant areas, such as borders. We found that these images were meaningless to humans, yet models can still classify them with high confidence, says Brandon Carter, MIT Computer Science and Artificial Intelligence Laboratory PhD student and lead author on a paper about the research.

Deep-image classifiers are widely used. In addition to medical diagnosis and boosting autonomous vehicle technology, there are use cases in security, gaming, and even an app that tells you if something is or isnt a hot dog, because sometimes we need reassurance. The tech in discussion works by processing individual pixels from tons of pre-labeled images for the network to learn.

Image classification is hard, because machine-learning models have the ability to latch onto these nonsensical subtle signals. Then, when image classifiers are trained on datasets such as ImageNet, they can make seemingly reliable predictions based on those signals.

Although these nonsensical signals can lead to model fragility in the real world, the signals are actually valid in the datasets, meaning overinterpretation cant be diagnosed using typical evaluation methods based on that accuracy.

To find the rationale for the model's prediction on a particular input, the methods in the present study start with the full image and repeatedly ask, what can I remove from this image? Essentially, it keeps covering up the image, until youre left with the smallest piece that still makes a confident decision.

To that end, it could also be possible to use these methods as a type of validation criteria. For example, if you have an autonomously driving car that uses a trained machine-learning method for recognizing stop signs, you could test that method by identifying the smallest input subset that constitutes a stop sign. If that consists of a tree branch, a particular time of day, or something that's not a stop sign, you could be concerned that the car might come to a stop at a place it's not supposed to.

While it may seem that the model is the likely culprit here, the datasets are more likely to blame. There's the question of how we can modify the datasets in a way that would enable models to be trained to more closely mimic how a human would think about classifying images and therefore, hopefully, generalize better in these real-world scenarios, like autonomous driving and medical diagnosis, so that the models don't have this nonsensical behavior, says Carter.

This may mean creating datasets in more controlled environments. Currently, its just pictures that are extracted from public domains that are then classified. But if you want to do object identification, for example, it might be necessary to train models with objects with an uninformative background.

This work was supported by Schmidt Futures and the National Institutes of Health. Carter wrote the paper alongside Siddhartha Jain and Jonas Mueller, scientists at Amazon, and MIT Professor David Gifford. They are presenting the work at the 2021 Conference on Neural Information Processing Systems.

Read more from the original source:
Nonsense can make sense to machine-learning models - MIT News

This past year has been a strange one in many respects: an ongoing pandemic, inflation, supply chain woes, uncertain plans for returning to the office, and worrying unemployment levels followed by the Great Resignation. After the shock of 2020, anyone hoping for a calm 2021 had to have been disappointed.

Data management and digital transformation remained in flux amid the ups and downs. Due to the ongoing challenges of the COVID-19 pandemic, as well as trends that were already underway prior to 2021, this retrospective article has a variety of enterprise AI, machine learning and data developments to cover.

Automation was a buzzword in 2021, thanks in part to the advantages that tools like automation software and robotics provided companies. As workplaces adapted to COVID-19 safety protocols, AI-powered automation proved beneficial. Since March 2020, two-thirds of companies have accelerated their adoption of AI and automation, consultancy McKinsey & Company found, making it one of the top AL and data trends of 2021.

In particular, robotic process automation (RPA) gained traction in several sectors, where it was put to use for tasks like processing transactions and sending notifications. RPA-focused firms like UiPath and tech giants like Microsoft went in on RPA this year. RPA software revenue will be up nearly 20% in 2021, according to research firm Gartner.

But while the pandemic may have sped up enterprise automation adoption, it appears RPA tools have lasting power. For example, Research and Markets predicted the RPA market will have a compound annual growth rate of 31.5% from 2021 to 2026. If 2020 was a year of RPA investment, 2021 and beyond will see those investments going to scale.

Micro-automation is one of the next steps in this area, said Mark Palmer, senior vice president of data, analytics and data science products at TIBCO Software, an enterprise data company. Adaptive, incremental, dynamic learning techniques are growing fields of AI/ML that, when applied to the RPAs exhaust, can make observations on the fly, Palmer said. These dynamic learning technologies help business users see and act on aha moments and make smarter decisions.

Automation also played an increasingly critical role in hybrid workplace models. While the tech sector has long accepted remote and hybrid work arrangements, other industries now embrace these models, as well. Automation tools can help offsite employees work efficiently and securely -- for example, by providing technical or HR support, security threat monitoring, and integrations with cloud-based services and software.

However, remote and hybrid workers do represent a potential pain point in one area: cybersecurity. With more employees working outside the corporate network, even if for only part of the work week, IT professionals must monitor more equipment for potential vulnerabilities.

The hybrid workforce influenced data trends in 2021. The wider distribution of IT infrastructure, along with increasing adoption of cloud-based services and software, added new layers of concerns about data storage and security. In addition, the surge in cyberattacks during the pandemic represented a substantial threat to enterprise data security. As organizations generate, store and use ever-greater amounts of data, an IT focus on cybersecurity is only going to become increasingly vital.

All together, these developments point to an overarching enterprise AI, ML and data trend for 2021: digital transformation. Spending on digital transformation is expected to hit $1.8 trillion in 2022, according to Statistica, which illustrates that organizations are willing to invest in this area.

As companies realize the value of data and the potential of machine learning in their operations, they also recognize the limitations posed by their legacy systems and outdated processes. The pandemic spurred many organizations to either launch or elevate digital transformation strategies, and those strategies will likely continue throughout 2022.

How did the AI, ML and data trends of 2021 change the way you work? Tell us in the comments below.

Original post:
Revisit Top AI, Machine Learning And Data Trends Of 2021 - ITPro Today

Machine learning is being utilized in service businesses to run standard, routine, repeatable parts of processes. During the recent OPEX Summer virtual conference, the daily sessions were filled with service companies presenting their approach to using machines to run the core business processes that are executed a dozen to a hundred times a day.

Manufacturing organizations can take a lesson from this approach. As we discussed in our earlier article, by leveraging a mixed-initiative approach and combining the best of Black Belt process expertise and machine learning systems, we can operationalize machine learning in a meaningful way and drive digital transformation into the manufacturing operation.

Machine algorithms are good at running repeatable processesthose that do not require human judgement to accomplish. However, the experts are still required to handle the edge cases, those that are non-standard and require some human intelligence to interpret and resolve. Edge cases in manufacturing involve non-routine things that happen infrequently and, on the surface, do not appear to be repeatable.

Some of these are extremely rare changes such as starting new production lines, qualifying next-generation equipment, replacement of outdated machinery, catastrophic equipment failure, etc. Other edge cases arise more frequently, such as when producing new productson restoration from failure and maintenance activitiesor when new operators are onboarded. In either case, the edge cases require some human intervention to resolve, re-optimize the process and bring it back to a stable state.

Getting machine-learning-based systems to handle edge cases is complex for several reasons:

Providing enough data to train a machine-learning-based approach requires experts to manually capture all actions used to manage the edge-case event and furthermore link these actions to the outcomes. This is problematic in manufacturing environments, where people are busy. Their value is not usually associated with data-entry tasks, but in units of output produced. Asking a person to manually input responses about an event that they have been busy recovering from is not likely to produce a quality dataset of responses.

In order to overcome these challenges, we require non-intrusive but continuous capture of actions and outcomes associated with an edge case event. There are several intelligent products out there with potential to bridge the gap. These include wearable technologies, as well as passive and intelligent interfaces. Google Glass is an example of the class of intelligent wearables that could be employed to bridge the gap. However, in this case, as opposed to providing real time assistance to the wearer to handle the edge case, we instead use the device to capture data, actions, and outcomes about edge cases. Similarly, we could also use an interactive and passive interface similar to the contact tracing approach adapted by Apple and Google. This has been used to enable a Bluetooth mesh network to trade data about Covid positive interactions without sharing privacy information, and can be repurposed for the factory floor to trace and record data tags when an edge case response is in process.

In addition to the non-intrusive capture of data, actions and outcomes, we also need advances in machine learning to be able to leverage this data to train models that can start to handle edge cases. An interesting area of research in machine learning is apprenticeship learning. The idea behind this is that the ML agent behaves like an apprenticeobserving the actions taken by the expert, and learning to mimic them to accomplish the appropriate task. These ideas have primarily been explored in robotics, where human experts are used to teach a robot agent how to take certain physical actions.

The underlying learning algorithms use inverse reinforcement learningwhere the model needs to estimate the objective an expert is trying to achieve from observing their actions, and then try to optimize it when it tries to accomplish a similar task. Recent applications of this approach have been shown to work in gaming environments (e.g. Atari game play) as well as in real-world settings such as helicopter control and animation. Adapting these approaches to the manufacturing environment would allow the ML agent to learn about actions needed to handle edge cases by observation.

The labor pinch that is the current reality will not abate for the remainder of this decade and into the next decade. Asking workers, of whom there is an ever-dwindling pool, to take time away from recovering from an event as fast as possible to enter data is a losing proposition. As the Great Resignation continues, the pressure on manufacturers will increase, as will turnover and demands for training as people filter through organizations in search of their ideal work situation.

As the available workforce dwindles, the machine needs to be able to absorb more and more of the edge content into the machine paradigm. Through a wearable monitoring product, passive tracking and inverse reinforcement-based learning approaches, the person can teach the machine about edge cases, which the machine can use to expand the understanding of the elements of response to edge cases that are routine, picking out elements that are repeatable even though edge cases don't happen every day.

As we march forward into the future, there will be population shrinkage. It is already happening in many countries. The portion of that future population that is willing to work in manufacturing will be a subset of a subset of a dwindling population, yet our demand for products seems to be increasing. Technology tools need to be assembled in such a way to bridge the gap.

The current state of manufacturing has several challenges to achieve the vision of machine- directed operations, with the digital aide concept at work. The economics of making the technology leap will change as the availability of cheap labor tightens. Many organizations have struggled for years to staff their operations, causing production outages and idle time, which is costly as the investment is underutilized. Additional challenges surround the comfort level of leaders with technology, ability to understand the potential for technology to solve their particular problems and patience as the technology approaches are put together into a seamless integration.

Manual data entry is a non-starter on the journey to enhancing the machines ability to learn the edge cases. Active monitoring tools that provide the data without the human having to stop their work on the edge case is the solution to achieve a learning machine. The imperative for the next decade is to set up the machine to learn from humans and absorb more of the edge cases by revealing the underlying routines and absorbing those routines in the library of Golden Runs.

Deepak Turagais senior vice president of data science at Oden Technologies, an industrial IoT company focused on using AI to monitor, optimize and control manufacturing processes. He has a background in academic and industry research specializing in using machine learning based tools to extract insights from streaming and real-time data. He is also an adjunct professor at Columbia University, and teaches a course on this topic every spring.

James Wellsisprincipal consultant at Quality in Practice, a consulting and training practice specializing in continuous improvement programs, and specializes in quality fundamentals, including the application of digital solutions to common manufacturing challenges. He has led quality and continuous improvement organizations for over 20 years at various manufacturing companies. Wells is a certified master Black Belt and certified lean specialist.

View original post here:
High-Level Machine Learning: What Will It Take? - IndustryWeek

MIT researchers have developed a deep learning model that can rapidly predict the likely 3D shapes of a molecule given a 2D graph of its structure. This technique could accelerate drug discovery. Credit: Courtesy of the researchers, edited by MIT News

A deep learning model rapidly predicts the 3D shapes of drug-like molecules, which could accelerate the process of discovering new medicines.

In their quest to discover effective new medicines, scientists search for drug-like molecules that can attach to disease-causing proteins and change their functionality. It is crucial that they know the 3D shape of a molecule to understand how it will attach to specific surfaces of the protein.

But a single molecule can fold in thousands of different ways, so solving that puzzle experimentally is a time-consuming and expensive process akin to searching for a needle in a molecular haystack.

MIT researchers are using machine learning to streamline this complex task. They have created a deep learning model that predicts the 3D shapes of a molecule solely based on a graph in 2D of its molecular structure. Molecules are typically represented as small graphs.

Their system, GeoMol, processes molecules in only seconds and performs better than other machine learning models, including some commercial methods. GeoMol could help pharmaceutical companies accelerate the drug discovery process by narrowing down the number of molecules they need to test in lab experiments, says Octavian-Eugen Ganea, a postdoc in the Computer Science and Artificial Intelligence Laboratory (CSAIL) and co-lead author of the paper.

When you are thinking about how these structures move in 3D space, there are really only certain parts of the molecule that are actually flexible, these rotatable bonds. One of the key innovations of our work is that we think about modeling the conformational flexibility like a chemical engineer would. It is really about trying to predict the potential distribution of rotatable bonds in the structure, says Lagnajit Pattanaik, a graduate student in the Department of Chemical Engineering and co-lead author of the paper.

Other authors include Connor W. Coley, the Henri Slezynger Career Development Assistant Professor of Chemical Engineering; Regina Barzilay, the School of Engineering Distinguished Professor for AI and Health in CSAIL; Klavs F. Jensen, the Warren K. Lewis Professor of Chemical Engineering; William H. Green, the Hoyt C. Hottel Professor in Chemical Engineering; and senior author Tommi S. Jaakkola, the Thomas Siebel Professor of Electrical Engineering in CSAIL and a member of the Institute for Data, Systems, and Society. The research will be presented this week at the Conference on Neural Information Processing Systems.

In a molecular graph, a molecules individual atoms are represented as nodes and the chemical bonds that connect them are edges.

GeoMol leverages a recent tool in deep learning called a message passing neural network, which is specifically designed to operate on graphs. The researchers adapted a message passing neural network to predict specific elements of molecular geometry.

Given a molecular graph, GeoMol initially predicts the lengths of the chemical bonds between atoms and the angles of those individual bonds. The way the atoms are arranged and connected determines which bonds can rotate.

GeoMol then predicts the structure of each atoms local neighborhood individually and assembles neighboring pairs of rotatable bonds by computing the torsion angles and then aligning them. A torsion angle determines the motion of three segments that are connected, in this case, three chemical bonds that connect four atoms.

Here, the rotatable bonds can take a huge range of possible values. So, the use of these message passing neural networks allows us to capture a lot of the local and global environments that influences that prediction. The rotatable bond can take multiple values, and we want our prediction to be able to reflect that underlying distribution, Pattanaik says.

One major challenge to predicting the 3D structure of molecules is to model chirality. A chiral molecule cant be superimposed on its mirror image, like a pair of hands (no matter how you rotate your hands, there is no way their features exactly line up). If a molecule is chiral, its mirror image wont interact with the environment in the same way.

This could cause medicines to interact with proteins incorrectly, which could result in dangerous side effects. Current machine learning methods often involve a long, complex optimization process to ensure chirality is correctly identified, Ganea says.

Because GeoMol determines the 3D structure of each bond individually, it explicitly defines chirality during the prediction process, eliminating the need for optimization after-the-fact.

After performing these predictions, GeoMol outputs a set of likely 3D structures for the molecule.

What we can do now is take our model and connect it end-to-end with a model that predicts this attachment to specific protein surfaces. Our model is not a separate pipeline. It is very easy to integrate with other deep learning models, Ganea says.

The researchers tested their model using a dataset of molecules and the likely 3D shapes they could take, which was developed by Rafael Gomez-Bombarelli, the Jeffrey Cheah Career Development Chair in Engineering, and graduate student Simon Axelrod.

They evaluated how many of these likely 3D structures their model was able to capture, in comparison to machine learning models and other methods.

In nearly all instances, GeoMol outperformed the other models on all tested metrics.

We found that our model is super-fast, which was really exciting to see. And importantly, as you add more rotatable bonds, you expect these algorithms to slow down significantly. But we didnt really see that. The speed scales nicely with the number of rotatable bonds, which is promising for using these types of models down the line, especially for applications where you are trying to quickly predict the 3D structures inside these proteins, Pattanaik says.

In the future, the researchers hope to apply GeoMol to the area of high-throughput virtual screening, using the model to determine small molecule structures that would interact with a specific protein. They also want to keep refining GeoMol with additional training data so it can more effectively predict the structure of long molecules with many flexible bonds.

Conformational analysis is a key component of numerous tasks in computer-aided drug design, and an important component in advancing machine learning approaches in drug discovery, says Pat Walters, senior vice president of computation at Relay Therapeutics, who was not involved in this research. Im excited by continuing advances in the field and thank MIT for contributing to broader learnings in this area.

Reference: GeoMol: Torsional Geometric Generation of Molecular 3D Conformer Ensembles by Octavian-Eugen Ganea, Lagnajit Pattanaik, Connor W. Coley, Regina Barzilay, Klavs F. Jensen, William H. Green and Tommi S. Jaakkola, 8 June 2021, Physics > Chemical Physics.arXiv:2106.07802

This research was funded by the Machine Learning for Pharmaceutical Discovery and Synthesis consortium.

Read the original here:
New Deep Learning Model Could Accelerate the Process of Discovering New Medicines - SciTechDaily

AutoML Market From $346. 2 million in 2020, the automated machine learning market is predicted to reach $14,830. 8 million by 2030, demonstrating a CAGR of 45. 6% from 2020 to 2030.

New York, Dec. 16, 2021 (GLOBE NEWSWIRE) -- Reportlinker.com announces the release of the report "AutoML Market" - https://www.reportlinker.com/p06191010/?utm_source=GNW The major factors driving the market are the burgeoning requirement for efficient fraud detection solutions, soaring demand for personalized product recommendations, and increasing need for predictive lead scoring.

The COVID-19 pandemic has contributed significantly to the evolution of digital business models, with many healthcare companies adopting machine-learning-enabled chatbots to enable the contactless screening of COVID-19 symptoms. Moreover, Clevy.io, which is a France-based start-up, and Amazon Web Services (AWS) have launched a chatbot for making the process of finding official government communications about the COVID-19 infection easy. Thus, the pandemic has positively impacted the market.

The service category, under the offering segment, is predicted to demonstrate the faster growth in the coming years. This is credited to the burgeoning requirement for implementation and integration, consulting, and maintenance services, as they assist in enhancing business productivity and augmenting coding activities. Additionally, these services aid in automating workflows, which, in turn, enables the mechanization of complex operations.

The cloud category dominated the AutoML market, within the deployment type segment, in the past. Moreover, this category is predicted to grow rapidly in the forthcoming years on account of the flexibility and scalability provided by cloud-based automated machine learning (AutoML) solutions.

Geographically, North America held the largest share in the past, and this trend is expected to continue in the coming years. This is credited to the soaring venture capital funding by artificial intelligence (AI) companies for research and development (R&D), in order to advance AutoML.

Asia-Pacific (APAC) is predicted to be the fastest-growing region in the market in the forthcoming years. This is ascribed to the growing information technology (IT) investments and increasing fintech adoption in the region. In addition, the growing government focus on incorporating AI in multiple verticals is supporting the advance of the market in the region.

For instance, in October 2021, Hivecell, which is an edge as a service company, entered into a partnership with DataRobot Inc. for solving bigger challenges and hurdles at the edge, by processing various ML models on site and outside the data closet. By incorporating the two solutions, businesses can make data-driven decisions more efficiently.

The major players in the AutoML market are DataRobot Inc., dotData Inc., H2O.ai Inc., Amazon Web Services Inc., Big Squid Inc., Microsoft Corporation, Determined.ai Inc., SAS Institute Inc., Squark, and EdgeVerve Systems Limited.Read the full report: https://www.reportlinker.com/p06191010/?utm_source=GNW

About ReportlinkerReportLinker is an award-winning market research solution. Reportlinker finds and organizes the latest industry data so you get all the market research you need - instantly, in one place.

__________________________

Story continues

Continued here:
The automated machine learning market is predicted to reach $14,830.8 million by 2030, demonstrating a CAGR of 45.6% from 2020 to 2030 - Yahoo Finance

WealthTech(Technology) companies have rapidly spawnedinrecent years. Cutting-edgetechnologies are making their wayinto almost allindustries from manufacturing to logistics to financial services.

Within financial services,technologies such as data analytics, ArtificialIntelligence, Machine Learning among others are leading the wayinchanging business processes with faster turnaround time and superior customer experience.See Zee Business Live TV Streaming Below:

Astechnology evolves, business models must be changed to remain relevant. Thewealthmanagementsectorisalso notinsulated from this phenomenon!

Ankur Maheshwari CEO-Wealth, Equirus decodes the impact of newtechnology advancementsinthewealthmanagementindustry:

Wealthtechupscalingthewealthmanagementspace

Wealthtechaids companiesindelivering a more convenient, hassle-free and engaging experience to clients at a relatively low cost.

The adoption of new-agetechnologies such as big data analytics, ArtificialIntelligence (AI), and Machine Learning (ML) are helpingwealthmanagementcompanies stay ahead of the curveinthe new age ofinvesting.

While the adoption of advancedtechnologies has been underway for quite some time, the pandemic has rapidlyincreased the pace of the adoption oftechnology.

New ageinvestors and the young population are usingtechnologyina big way. Thisisevident from the fact that the total digital transactionsinIndia have grown from 14.59 billioninFY18 to43.71 billioninFY21 as reported by the RBI.

According to a report released by ACI Worldwide Globally, more than 70.3 billion real-time transactions were processedinthe year 2020, withIndia at the top spot with more than 25 billion real-time payment transactions.

Thisindicates the rising use oftechnology globally andinIndia within the financial servicesindustry.

There are various areas wheretechnology has had a significant impact on client experience and offerings ofwealthmanagementcompanies.

Client Meetings andInteractions

Inthe old days,wealthmanagers would physically meet theinvestors to discuss theirwealthmanagementrequirements. However, recently we see that a lot ofinvestors are demanding more digital touchpointswhichoffer more convenience.

Video calling and shared desktop features have been rapidly adopted by bothinvestors andwealthmanagers to provide a seamless experience.

24*7 digital touchpoints available

Technology has also enabled companies to provide cost-effective digital touchpoint solutions to clients that enable easier and faster access to portfolio updates, various reports such as capital gains reports, and holding statements and enable ease of doing transactions.

Features such as Chatbots and WhatsApp-enabled touchpoints are helpingindelivering a high-end client experienceina quick turnaround time.

Portfolio analytics and reporting

Data analytics has not only augmented the waywealthmanagers analyseinvestors portfolios but have also reduced time spent bywealthmanagers on spreadsheets.

WealthTechalso offers deeperinsightsinto the portfolioswhichassistwealthmanagersinproviding a more comprehensive and customized offering toinvestorswhichmatch their expectations and risk appetite.

ArtificialIntelligence and Machine Learningtechnologies combined with big data analytics are disruptingwealthmanagementspaceina big way. Robo-advisory and quant-based product offerings are making strong headwayinto thisspace.

Ease of process and documentation

Inthe earlier days, documentation and KYC process used to be a bottleneck with processing time goinginto several days as wellinsome cases. Storage of documentsisalso challenging as this requires safe storagespaceand documents are prone to damage and/or being misplaced.

With the advancementintechnologies, we are now moving towards a fully digital and/or phy-gital mode of operations. Whileinvestinginsome products like mutual funds the processiscompletely digital for other products like PMS, AIF, structures, etc. the processes are moving towards phy-gital mode.

The use of Aadhar based digital signature and video KYC have made it possible to reduce the overall processing time significantly!

Summing up:

A shift towards holistic offerings rather than product-based offering

Theincreasing young populationiscominginto the workforce and thereby creating a shiftinfocus towards new-ageinvestors.

These new-ageinvestors are not onlytech-savvy and early adopters oftechnology but are also demanding moreinterms of offerings.

With easy access toinformation and growing awareness,investors are looking for holistic offerings rather than merely product-based offeringswhichencompass all theirwealthmanagementneeds.

Incumbentsinthewealthmanagementspaceshould, if they havent already,incorporatetechnology as anintegral part of their client offering to stay relevant.

Forincumbents, it may prove to be cheaper and faster to getinto the tie-ups, partnerships, or acquire new agetechnology companies to quickly come up the curve rather than buildingin-housetechnology solutions.

As the adage goes, the only constantinlifeischange;technologyisa change for thewealthmanagementdomain that needs to be embraced!

(Disclaimer: The views/suggestions/advice expressed hereinthis article are solely byinvestment experts. Zee Business suggests its readers to consult with theirinvestment advisers before making any financial decision.)

Read more here:
From AI to Machine Learning, 4 ways in which technology is upscaling wealth management space - Zee Business

Given the growing ubiquity of machine learning and artificial intelligence in healthcare settings, it's become increasingly important to meet patient needs and engage users.

And as panelists noted during a HIMSS Machine Learning and AI for Healthcare Forum session this week, designing technology with the user in mind is a vital way to ensure tools become an integral part of workflow.

"Big Tech has stumbled somewhat" in this regard, said Bill Fox, healthcare and life sciences lead at SambaNova Systems. "The patients, the providers they don't really care that much about the technology, how cool it is, what it can do from a technological standpoint.

"It really has to work for them," Fox added.

Jai Nahar, a pediatric cardiologist at Children's National Hospital, agreed, stressing the importance of human-centered AI design in healthcare delivery.

"Whenever we're trying to roll out a productive solution that incorporates AI," he said, "right from the designing [stage] of the product or service itself, the patients should be involved."

That inclusion should also expand to provider users too, he said: "Before rolling out any product or service, we should involve physicians or clinicians who are going to use the technology."

The panel, moderated by Rebekah Angove, vice president of evaluation and patient experience at the Patient Advocate Foundation, noted that AI is already affecting patients both directly and indirectly.

In ideal scenarios, for example, it's empowering doctors to spend more time with individuals. "There's going tobe a human in the loop for a very long time," said Fox.

"We can power the clinician with better information from a much larger data set," he continued. AI is also enabling screening tools and patient access, said the experts.

"There are many things that work in the background that impact [patient] lives and experience already," said Piyush Mathur, staff anesthesiologist and critical care physician at the Cleveland Clinic.

At the same time, the panel pointed to the role clinicians can play in building patient trust around artificial intelligence and machine learning technology.

Nahar said that as a provider, he considers several questions when using an AI-powered tool for his patient. "Is the technology really needed for this patient to solve this problem?" he said he asks himself. "How will it improve the care that I deliver to the patient? Is it something reliable?"

"Those are the points, as a physician, I would like to know," he said.

Mathur also raised the issue of educating clinicians about AI. "We have to understand it a little bit better to be able to translate that science to the patients in their own language," he said. "We have to be the guardians of making sure that we're providing the right data for the patient."

The panelists discussed the problem of bias, about which patients may have concerns and rightly so.

"There are multiple entry points at which bias can be introduced," said Nahar.

During the design process, he said, multiple stakeholders need to be involved to closely consider where bias could be coming from and how it can be mitigated.

As panelists have pointed out at other sessions, he also emphasized the importance of evaluating tools in an ongoing process.

Developers and users should be asking themselves, "How can we improve and make it better?" he said.

Overall, said Nahar, best practices and guidances need to be established to better implement and operationalize AI from the patient perspective and provider perspective.

The onus is "upon us to make sure we use this technology in the correct way to improve care for our patients," added Mathur.

Kat Jercich is senior editor of Healthcare IT News.Twitter: @kjercichEmail: kjercich@himss.orgHealthcare IT News is a HIMSS Media publication.

Read the original:
Human-centered AI can improve the patient experience - Healthcare IT News

newsletters, editors-pick-list,

Artificial intelligence is being used to detect and predict depression in people in a University of Newcastle research project that aims to improve quality of life. Associate Professor Raymond Chiong's research team has developed machine-learning models that "detect signs of depression using social media posts with over 98 per cent accuracy". "We have used machine learning to analyse social media posts such as tweets, journal entries, as well as environmental factors such as demographic, social and economic information about a person," Dr Chiong said. This was done to detect if people were suffering from depression and to "predict their likelihood of suffering from depression in the future". Dr Chiong said early detection of depression and poor mental health can "prevent self-harm, relapse or suicide, as well as improve the quality of life" of those affected. "More than four million Australians suffer from depression every year and over 3000 die from suicide, with depression being a major risk factor," he said. People often use social media to "express their feelings" and this can "identify multiple aspects of psychological concerns and human behaviour". The next stage of the team's research will involve "detecting signs of depression by analysing physiological data collected from different kinds of devices". "This should allow us to make more reliable and actionable predictions/detections of a person's mental health, even when all data sources are not available," he said. "Data from wearable devices such as activity measurements, heart rate and sleeping patterns can be used for behaviour and physiological monitoring. "By combining and analysing data from these sources, we can potentially get a very good picture of a person's mental health." The goal is to make such tools available on a smartphone application, which will allow people to regularly monitor their mental health and seek help in the early stages of depression. "Such an app will also build the ability of mental health and wellbeing providers to integrate digital technologies when monitoring their patients, by giving them a source of regular updates about the mental health status of their patients," he said. "We want to use artificial intelligence and machine learning to develop tools that can detect signs of depression by utilising data from things we use on a regular basis, such as social media posts, or data from smartwatches or fitness devices." The research team aims to develop smartphone apps that can be used by mental health professionals to better monitor their patients and help them provide more effective treatment. The overarching goal of the research is to "improve quality of life". "Depression can seriously impact one's enjoyment of life. It does not discriminate - anyone can suffer from it," Dr Chiong said. "To live a high quality of life, one needs to be in good mental health. Good mental health helps people deal with environmental stressors, such as loss of a job or partner, illness and many other challenges in life." The technology involved can help people monitor how well they are coping in challenging circumstances. This can encourage them to seek help from family, friends and professionals in the early stages of ailing mental health. By doing so, professionals could help people prone to depression and other mental illnesses well before the situation becomes risky. "They could also use this technology to get more information about their patients, in addition to what they can glean during consultation," he said. This makes early interventions possible and "reduces the likelihood of self-harm or suicide attempts". Depending on funding, the team plans to work on integrating people's health data from smart-fitness devices, such as heart rate, sleeping patterns and physical activity. The intention is to work with Hunter New England mental health professionals on this stage of the research. "Following this, our goal is to develop a smartphone app that can not only be used by clinical practitioners, but also everyday individuals to monitor their mental health status in real time." He said machine learning models had shown "great potential in terms of learning from training data and making highly accurate predictions". "For example, the application of machine learning/deep learning for image recognition is a major success story," he said. Studies have shown that machine learning had "enormous potential in the field of mental health as well". "The fact that we were able to obtain more than 98 per cent accuracy in detecting signs of ill mental health demonstrates that there is great potential for machine learning in this field." However, he said the technology does face challenges before it can be applied in real-world scenarios. "Some mobile apps have been developed that use machine learning to provide customised physical or other activities for their users, with the goal of helping them stay in good mental health," he said. "However, our proposed app will be one of the first that allows users to monitor their mental health status in real time, by analysing their social media posts and health measurements." Clinical practitioners could use this app to monitor their patients, but convincing them to use the technology will be one of the challenges.

/images/transform/v1/crop/frm/3AijacentBN9GedHCvcASxG/cf2280ff-31ca-4da2-bbb1-672ee0fdc28e.jpg/r1431_550_4993_2563_w1200_h678_fmax.jpg

December 19 2021 - 4:30PM

Detection: Dr Raymond Chiong said "we can potentially get a very good picture of a person's mental health" with artificial intelligence. Picture: Simone De Peak

Artificial intelligence is being used to detect and predict depression in people in a University of Newcastle research project that aims to improve quality of life.

Associate Professor Raymond Chiong's research team has developed machine-learning models that "detect signs of depression using social media posts with over 98 per cent accuracy".

"We have used machine learning to analyse social media posts such as tweets, journal entries, as well as environmental factors such as demographic, social and economic information about a person," Dr Chiong said.

This was done to detect if people were suffering from depression and to "predict their likelihood of suffering from depression in the future".

Dr Chiong said early detection of depression and poor mental health can "prevent self-harm, relapse or suicide, as well as improve the quality of life" of those affected.

"More than four million Australians suffer from depression every year and over 3000 die from suicide, with depression being a major risk factor," he said.

People often use social media to "express their feelings" and this can "identify multiple aspects of psychological concerns and human behaviour".

The next stage of the team's research will involve "detecting signs of depression by analysing physiological data collected from different kinds of devices".

"This should allow us to make more reliable and actionable predictions/detections of a person's mental health, even when all data sources are not available," he said.

"Data from wearable devices such as activity measurements, heart rate and sleeping patterns can be used for behaviour and physiological monitoring.

"By combining and analysing data from these sources, we can potentially get a very good picture of a person's mental health."

The goal is to make such tools available on a smartphone application, which will allow people to regularly monitor their mental health and seek help in the early stages of depression.

"Such an app will also build the ability of mental health and wellbeing providers to integrate digital technologies when monitoring their patients, by giving them a source of regular updates about the mental health status of their patients," he said.

"We want to use artificial intelligence and machine learning to develop tools that can detect signs of depression by utilising data from things we use on a regular basis, such as social media posts, or data from smartwatches or fitness devices."

The research team aims to develop smartphone apps that can be used by mental health professionals to better monitor their patients and help them provide more effective treatment.

The overarching goal of the research is to "improve quality of life".

"Depression can seriously impact one's enjoyment of life. It does not discriminate - anyone can suffer from it," Dr Chiong said.

"To live a high quality of life, one needs to be in good mental health. Good mental health helps people deal with environmental stressors, such as loss of a job or partner, illness and many other challenges in life."

The technology involved can help people monitor how well they are coping in challenging circumstances.

This can encourage them to seek help from family, friends and professionals in the early stages of ailing mental health.

By doing so, professionals could help people prone to depression and other mental illnesses well before the situation becomes risky.

"They could also use this technology to get more information about their patients, in addition to what they can glean during consultation," he said.

This makes early interventions possible and "reduces the likelihood of self-harm or suicide attempts".

Depending on funding, the team plans to work on integrating people's health data from smart-fitness devices, such as heart rate, sleeping patterns and physical activity.

The intention is to work with Hunter New England mental health professionals on this stage of the research.

"Following this, our goal is to develop a smartphone app that can not only be used by clinical practitioners, but also everyday individuals to monitor their mental health status in real time."

He said machine learning models had shown "great potential in terms of learning from training data and making highly accurate predictions".

"For example, the application of machine learning/deep learning for image recognition is a major success story," he said.

Studies have shown that machine learning had "enormous potential in the field of mental health as well".

"The fact that we were able to obtain more than 98 per cent accuracy in detecting signs of ill mental health demonstrates that there is great potential for machine learning in this field."

However, he said the technology does face challenges before it can be applied in real-world scenarios.

"Some mobile apps have been developed that use machine learning to provide customised physical or other activities for their users, with the goal of helping them stay in good mental health," he said.

"However, our proposed app will be one of the first that allows users to monitor their mental health status in real time, by analysing their social media posts and health measurements."

Clinical practitioners could use this app to monitor their patients, but convincing them to use the technology will be one of the challenges.

Visit link:
Artificial intelligence and machine learning can detect and predict depression in University of Newcastle research - Newcastle Herald

By Columbia University School of Engineering and Applied ScienceDecember 12, 2021

Schematic of the bridging of the cold quantum world and high-temperature metal extraction with machine learning. Credit: Rodrigo Ortiz de la Morena and Jose A. Garrido Torres/Columbia Engineering

Method combines quantum mechanics with machine learning to accurately predict oxide reactions at high temperatures when no experimental data is available; could be used to design clean carbon-neutral processes for steel production and metal recycling.

Extracting metals from oxides at high temperatures is essential not only for producing metals such as steel but also for recycling. Because current extraction processes are very carbon-intensive, emitting large quantities of greenhouse gases, researchers have been exploring new approaches to developing greener processes. This work has been especially challenging to do in the lab because it requires costly reactors. Building and running computer simulations would be an alternative, but currently there is no computational method that can accurately predict oxide reactions at high temperatures when no experimental data is available.

A Columbia Engineering team reports that they have developed a new computation technique that, through combining quantum mechanics and machine learning, can accurately predict the reduction temperature of metal oxides to their base metals. Their approach is computationally as efficient as conventional calculations at zero temperature and, in their tests, more accurate than computationally demanding simulations of temperature effects using quantum chemistry methods. The study, led by Alexander Urban, assistant professor of chemical engineering, was published on December 1, 2021 by Nature Communications.

Decarbonizing the chemical industry is critical if we are to transition to a more sustainable future, but developing alternatives for established industrial processes is very cost-intensive and time-consuming, Urban said. A bottom-up computational process design that doesnt require initial experimental input would be an attractive alternative but has so far not been realized. This new study is, to our knowledge, the first time that a hybrid approach, combining computational calculations with AI, has been attempted for this application. And its the first demonstration that quantum-mechanics-based calculations can be used for the design of high-temperature processes.

The researchers knew that, at very low temperatures, quantum-mechanics-based calculations can accurately predict the energy that chemical reactions require or release. They augmented this zero-temperature theory with a machine-learning model that learned the temperature dependence from publicly available high-temperature measurements. They designed their approach, which focused on extracting metal at high temperatures, to also predict the change of the free energy with the temperature, whether it was high or low.

Free energy is a key quantity of thermodynamics and other temperature-dependent quantities can, in principle, be derived from it, said Jos A. Garrido Torres, the papers first author who was a postdoctoral fellow in Urbans lab and is now a research scientist at Princeton. So we expect that our approach will also be useful to predict, for example, melting temperatures and solubilities for the design of clean electrolytic metal extraction processes that are powered by renewable electric energy.

The future just got a little bit closer, said Nick Birbilis, Deputy Dean of the Australian National University College of Engineering and Computer Science and an expert for materials design with a focus on corrosion durability, who was not involved in the study. Much of the human effort and sunken capital over the past century has been in the development of materials that we use every day and that we rely on for our power, flight, and entertainment. Materials development is slow and costly, which makes machine learning a critical development for future materials design. In order for machine learning and AI to meet their potential, models must be mechanistically relevant and interpretable. This is precisely what the work of Urban and Garrido Torres demonstrates. Furthermore, the work takes a whole-of-system approach for one of the first times, linking atomistic simulations on one end engineering applications on the other via advanced algorithms.

The team is now working on extending the approach to other temperature-dependent materials properties, such as solubility, conductivity, and melting, that are needed to design electrolytic metal extraction processes that are carbon-free and powered by clean electric energy.

Reference: Augmenting zero-Kelvin quantum mechanics with machine learning for the prediction of chemical reactions at high temperatures by Jose Antonio Garrido Torres, Vahe Gharakhanyan, Nongnuch Artrith, Tobias Hoffmann Eegholm and Alexander Urban, 1 December 2021, Nature Communications.DOI: 10.1038/s41467-021-27154-2

View original post here:
Quantum Mechanics and Machine Learning Used To Accurately Predict Chemical Reactions at High Temperatures - SciTechDaily