Mediaboss Marketing

It has been observed by many that we are at the dawn of the next industrial revolution: The Artificial Intelligence (AI) revolution. The benefits delivered by this intelligence revolution will be many: in medicine, improved diagnostics and precision treatment, better weather forecasting, and self-driving vehicles to name a few. However, one of the costs of this revolution is going to be increased electrical consumption by the data centers that will power it. Data center power usage is projected to double over the next 10 years and is on track to consume 11% of worldwide electricity by 2030. Beyond AI adoption, other drivers of this trend are the movement to the cloud and increased power usage of CPUs, GPUs and other server components, which are becoming more powerful and smart.

AIs two basic elements, training and inference, each consume power differently. Training involves computationally intensive matrix operations over very large data sets, often measured in terabytes to petabytes. Examples of these data sets can range from online sales data to captured video feeds to ultra-high-resolution images of tumors. AI inference is computationally much lighter in nature, but can run indefinitely as a service, which draws a lot of power when hit with a large number of requests. Think of a facial recognition application for security in an office building. It runs continuously but would stress the compute and storage resources at 8:00am and again at 5:00pm as people come and go to work.

However, getting a good handle on power usage in AI is difficult. Energy consumption is not part of standard metrics tracked by job schedulers and while it can be set up, it is complicated and vendor dependent. This means that most users are flying blind when it comes to energy usage.

To map out AI energy requirements, Dr. Miro Hodak led a team of Lenovo engineers and researchers, which looked at the energy cost of an often-used AI workload. The study, Towards Power Efficiency in Deep Learning on Data Center Hardware, (registration required) was recently presented at the 2019 IEEE International Conference on Big Data and was published in the conference proceedings. This work looks at the energy cost of training ResNet50 neural net with ImageNet dataset of more than 1.3 million images on a Lenovo ThinkSystem SR670 server equipped with 4 Nvidia V100 GPUs. AC data from the servers power supply, indicates that 6.3 kWh of energy, enough to power an average home for six hours, is needed to fully train this AI model. In practice, trainings like these are repeated multiple times to tune the resulting models, resulting in energy costs that are actually several times higher.

The study breaks down the total energy into its components as shown in Fig. 1. As expected, the bulk of the energy is consumed by the GPUs. However, given that the GPUs handle all of the computationally intensive parts, the 65% share of energy is lower than expected. This shows that simplistic estimates of AI energy costs using only GPU power are inaccurate and miss significant contributions from the rest of the system. Besides GPUs, CPU and memory account for almost quarter of the energy use and 9% of energy is spent on AC to DC power conversion (this is within line of 80 PLUS Platinum certification of SR670 PSUs).

The study also investigated ways to decrease energy cost by system tuning without changing the AI workload. We found that two types of system settings make most difference: UEFI settings and GPU OS-level settings. ThinkSystem servers provides four UEFI running modes: Favor Performance, Favor Energy, Maximum Performance and Minimum Power. As shown in Table 1, the last option is the best and provides up to 5% energy savings. On the GPU side, 16% of energy can be saved by capping V100 frequency to 1005 MHz as shown in Figure 2. Taking together, our study showed that system tunings can decrease energy usage by 22% while increasing runtime by 14%. Alternatively, if this runtime cost is unacceptable, a second set of tunings, which save 18% of energy while increasing time by only 4%, was also identified. This demonstrates that there is lot of space on system side for improvements in energy efficiency.

Energy usage in HPC has been a visible challenge for over a decade, and Lenovo has long been a leader in energy efficient computing. Whether through our innovative Neptune liquid-cooled system designs, or through Energy-Aware Runtime (EAR) software, a technology developed in collaboration with Barcelona Supercomputing Center (BSC). EAR analyzes user applications to find optimum CPU frequencies to run them at. For now, EAR is CPU-only, but investigations into extending it to GPUs are ongoing. Results of our study show that that is a very promising way to bring energy savings to both HPC and AI.

Enterprises are not used to grappling with the large power profiles that AI requires, the way HPC users have become accustomed. Scaling out these AI solutions will only make that problem more acute. The industry is beginning to respond. MLPerf, currently the leading collaborative project for AI performance evaluation, is preparing new specifications for power efficiency. For now, it is limited to inference workloads and will most likely be voluntary, but it represents a step in the right direction.

So, in order to enjoy those precise weather forecasts and self-driven cars, well need to solve the power challenges they create. Today, as the power profile of CPUs and GPUs surges ever upward, enterprise customers face a choice between three factors: system density (the number of servers in a rack), performance and energy efficiency. Indeed, many enterprises are accustomed to filling up rack after rack with low cost, adequately performing systems that have limited to no impact on the electric bill. Unfortunately, until the power dilemma is solved, those users must be content with choosing only two of those three factors.

View post:
Powering the Artificial Intelligence Revolution - HPCwire

According to a recent study, humans and artificial intelligence systems can perform better when both of them work together to tackle problems. The research was done by Eric Horvitz Chief scientist Microsoft, Ece Kamar the Microsoft Research principal researcher, and Bryan Wilder, a student at Harvard University and Microsoft Research intern.

It seems that Eric Horvitz first published the research paper. He was hired as Microsoft principal researcher back in the year 1993 and the company named him Microsoft Chief Scientist officer during March. He led the companys research programs from the year 2017 to 2020. The research paper was published earlier this month and it studies the performance of artificial intelligence teams and humans operating together on two PC vision projects namely breast cancer metastasis recognition and Galaxy categorization. With this proposed approach, the artificial intelligence (AI) model evaluates which tasks humans can perform best and what type of tasks AI systems can handle better.

In this approach, the learning procedure is developed to merge human contributions and machine predictions. The artificial intelligence systems work to tackle problems that can be difficult for humans while humans focus on solving issues that can be tough for AI systems to figure out. Basically, AI system predictions generated with lower accuracy levels are routed to human teams in this system. According to researchers, combined training of human and artificial intelligence systems can enhance the galaxy classification model for us. It can improve the performance of Galaxy Zoo with a 21 to 73% decrease in loss. This system can also deliver an up to 20% better performance for CAMELYON16.

The research paper states that the performance of machine learning in segregation overcomes the circumstances where human skills can add integral context, although human teams have their own restrictions including systematic biases. Researchers stated in the paper that they have developed methods focused on training the AI learning model to supplement human strengths. It also accounts for the expense of inquiring an expert. Human and AI system teamwork can take various forms but the researchers focused on settings where machines would decide which instances required human absorption and then merging human and machine judgments.

Horvitz, during the year 2007, worked on a policy to examine when human assistants should interfere in consumer conversations with computerized receptionist systems. The researchers also stated in the paper, Learning to Complement Humans, that they see opportunities of studying extra aspects of human-machine cooperation across various settings. While studying a different type of teamwork, Open Artificial Intelligence research experts have looked at machine assistants operating together in games such as hide and seek, and Quake 3.

Photo: Ipopba / Getty Images

Read next: Researchers Developed An Artificial Intelligence System That Can Transform Brain Signals Into Words

Excerpt from:
Humans And Artificial Intelligence Systems Perform Better Together: Microsoft Chief Scientist Eric Horvitz - Digital Information World

Among the myriad thematic exchange traded funds investors have to consider, artificial intelligence products are numerous and some are catching on with investors.

Count the ROBO Global Artificial Intelligence ETF THNQ, +0.40% as the latest member of the artificial intelligence ETF fray. HNQ, which debuted earlier this week, comes from a good gene pool as its stablemate,the Robo Global Robotics and Automation Index ETF ROBO, -0.32%, was the original and remains one of the largest robotics ETFs.

That's relevant because artificial intelligence and robotics are themes that frequently intersect with each other. Home to 72 stocks, the new THNQ follows the ROBO Global Artificial Intelligence Index.

Adding to the case for A.I., even with a new product such as THNQ, is that the technology has hundreds, if not thousands, of applications supporting its growth.

Companies developing AV technology are mainly relying on machine learning or deep learning, or both, according to IHS Markit. A major difference between machine learning and deep learning is that, while deep learning can automatically discover the feature to be used for classification in unsupervised exercises, machine learning requires these features to be labeled manually with more rigid rulesets. In contrast to machine learning, deep learning requires significant computing power and training data to deliver more accurate results.

Like its family ROBO, THNQ offers wide reach with exposure to 11 sub-groups. Those include big data, cloud computing, cognitive computing, e-commerce and other consumer angles and factory automation, among others. Of course, semiconductors are part of the THNQ fold, too.

The exploding use of AI is ushering in a new era of semiconductor architectures and computing platforms that can handle the accelerated processing requirements of an AI-driven world, according to ROBO Global. To tackle the challenge, semiconductor companies are creating new, more advanced AI chip engines using a whole new range of materials, equipment, and design methodologies.

While THNQ is a new ETF, investors may do well to not focus on that rather focus on the fact the AI boom is in its nascent stages.

Historically, the stock market tends to under-appreciate the scale of opportunity enjoyed by leading providers of new technologies during this phase of development, notes THNQ's issuer. This fact creates a remarkable opportunity for investors who understand the scope of the AI revolution, and who take action at a time when AI is disrupting industry as we know it and forcing us to rethink the world around us.

The new ETF charges 0.68% per year, or $68 on a $10,000 investment. That's inline with rival funds.

2020 Benzinga.com. Benzinga does not provide investment advice. All rights reserved.

Read more:
A New Way To Think About Artificial Intelligence With This ETF - MarketWatch

ARLINGTON, VA, May 19, 2020 (GLOBE NEWSWIRE) -- The artificial intelligence industry is rapidly growing, providing an opportunity to enhance government data security in the United States. Enabled Intelligence, Inc. and SourceAmerica have recently partnered to expand competitive integrative employment for professionals with disabilities.

Enabled Intelligence is an artificial intelligence company that provides highly secure and accurate data labeling services. SourceAmerica is a national nonprofit organization committed to providing employment opportunities for people with disabilities through its network of more than 600 community-based nonprofit agencies across the country. Together, they will work to recruit and train highly capable people with disabilities to join Enabled Intelligences growing tech workforce.

Enabled Intelligence is expanding its workforce to meet the U.S. governments rapidly increasing demand for secure high-quality data labeling to support artificial intelligence technology development. The Department of Defense, intelligence agencies and other federal programs are increasingly deploying emerging artificial intelligence technologies and accurately labeled data to train those systems. Enabled Intelligences workforce of highly-trained U.S. based employees provide the subject matter expertise and secure systems able to handle the government's most sensitive data.

We are honored to be working with SourceAmerica as we expand our integrated team including professionals with disabilities. People with disabilities are often overlooked as a resource but they are invaluable to us in their commitment to service and excellent labeling skills, explained Peter Kant, CEO of Enabled Intelligence.

SourceAmerica is pleased to partner with Enabled Intelligence to make an impact in the artificial intelligence industry, said Vince Loose, president and CEO of SourceAmerica. Professionals with disabilities will bring unique insights and talents to this relationship with Enabled Intelligence and their federal and commercial customers who are looking to enhance their capabilities in this area.

About Enabled Intelligence, Inc.Enabled Intelligence is a small company based in Arlington, Virginia providing sensitive and classified data labeling services for government and other critical artificial intelligence applications. The company is hyper focused on labeling accuracy and security employing a competitive integrated team of professionals including veterans, people with disabilities and other subject matter experts. Visitwww.enabledintelligence.net to learn more.

About SourceAmericaEstablished in 1974, SourceAmerica creates employment opportunities for a skilled and dedicated workforce of people with disabilities. SourceAmerica is the vital link between the federal government and private sector organizations that procure the products and services provided by this exceptional workforce via a network of more than 600 community-based nonprofits. Headquartered in Vienna, Virginia, SourceAmerica provides its nonprofit agency network with business development, contract management, legislative and regulatory assistance, communications and public relations materials, information technology support, engineering and technical assistance, and extensive professional training needed for successful nonprofit management. Visit SourceAmerica.org to learn more, or follow them onFacebook(@SourceAmerica),Twitter(@SourceAmericaUS) andLinkedIn(@SourceAmerica).

See the article here:
Enabled Intelligence, Inc. and SourceAmerica announce partnership to expand high-tech employment opportunities for people with disabilities -...

Dr D Narayana, Professor, AIML, D Arpna, S Peeyushi, S Samiksha, V Sanjay and P Sneha, Researchers, Great Learning discuss use of AI, ML which can boost process by identifying drugs having efficacy against COVID-19, bridging gap between thousands of repurposed drugs

COVID-19 pandemic has spread far and wide and has been different from the other pandemics of the last few decades. In India, where the first case was reported on January 30, 2020, till date there have been 1,01,139 confirmed cases of COVID-19 with 3,163 deaths. Many countries worldwide, including India are under lockdown to avoid the spread of the disease. However, to fight the disease effectively, the need of the hour is vaccines to combat the virus.

There are two broad categories of vaccines prophylactic and therapeutic. While prophylactic vaccines make a person immune to the virus, therapeutic ones are for making body fight against the virus which has already entered the body.

Many prophylactic vaccines are under trial world over but chances of those being mass produced and reaching India early on seem to be low. In India, due to low hospitals to population ratio, the focus should be on therapeutic vaccines to reduce the number of cases of hospitalisation.

Amongst therapeutic vaccines, repurposed drugs (using existing drugs for other diseases) should be our first line of attack against the pandemic. Other reasons for the focus on repurposed drugs are low chances of adverse reaction to the host (human) body as those drugs are already being used for treating other conditions. Also, these drugs can be used immediately and hence save many lives.

Repurposed drugs are being identified to interrupt or block different stages of the virus life cycle. Virus life cycle ranges from host cell penetration to self-replication inside the cell to exocytosis of new virions from the host cell.

For ease of understanding, we have categorised the repurposed drugs identified world over, as per the stage of the virus life cycle at which those are effective.

Virus entry blockers like camostat mesylate, a protease inhibitor, shown to inhibit TMPRSS2 (used for cleaving spike protein during virus entry). Its clinical trial for COVID-19 was started on April 3, 2020 (the drug is already licensed in Japan and South Korea for pancreatitis). The antimalarial drug, Hydroxychloroquine, can increase the endosomal pH required for virus-cell fusion and hence can potentially block the viral infection whereas another antimalarial drug, chloroquine phosphate, can target ACE2 cells. However, the study on these anti-malarial drugs in France led to no improvement in patients.

Virus replication blockers like Remedesivir and Favipiravir can interfere with RdRP (RNA dependent RNA polymerase) which is a viral generated protein responsible for intracellular sub-genomic RNA production. On May 1, 2020, Remedesivir was granted Emergency use Drug authorisation by US FDA whereas Favipiravir is emerging as one of the top drugs being recommended by CSIR (Council of Scientific and Industrial research), India. Ivermectin, a drug to treat broad spectrum parasitic infections, was studied by Australian researchers in-vitro and it was found that the drug was able to stop the virus replication. However, questions are being raised on toxicity of the dosage required.

Cytokine storm cytokine storms occur in viral infections when a large number of cytokines are produced. It is associated with multi-organ failure, which is frequently fatal. During infection from SARS-CoV-2, this cytokine storm is associated with increased levels of interleukins IL 2-2, IL-7 and other cytokines. A multi-centre, randomised controlled trial of tocilizumab (IL-6 receptor blockade, licensed for cytokine release syndrome), has been approved in patients with COVID-19 pneumonia and elevated IL-6 in China.

Potential natural drugs

Recently Indian governments CTRI (an arm of the Indian Council of Medical Research), has provided approval to conduct a randomised multicentre interventional clinical trial of a repurposed ayurvedic drug named as Zingivir-H. This drug, developed by Pankajakasthuri Herbal Research Foundation, an ayurvedic organisation from Kerala, is part of clinical practice for nearly 15 years for viral fever, acute viral bronchitis and contagious fever. It has been found to not have any side effects as per in-vitro experiments carried out at Rajiv Gandhi Centre for Biotechnology. It has seven ingredients including herbomineral and these ingredients are part of scientific manuscript. Additionally, studies have been carried out to check the efficacy of 64 naturally occurring flavonoids. Hesperidin, herbacetin, rhoifolin and pectolinarin were found to efficiently block the enzymatic activity of SARS-CoV 3CLpro.

As per latest statistics, the trial count of most popular allopathic drugs are as follows:

All the above drugs have been identified / shortlisted by researchers all over the world by using pre-existing drug repositories. Those drug repositories have been filtered / scanned to identify the ones with high affinity for virus proteins and hence leading to interruption of key activities of the virus during its life cycle.

Few open source repositories are : ReDO database which is maintained by AntiCancer Fund, Excelra Repurposed Drugs Database, CAS antiviral drugs dataset, DrugBank Database, the database of commercially available compounds for virtual screening known as ZINC, PubChem and ChEMBL dataset etc.

Sifting through thousands of these drug repos and coming up with the most effective drugs in itself is a time-consuming process. Artificial intelligence(AI) and machine learning(ML) can serve as a booster for this search by narrowing down the most effective drugs amongst the lot which can be further studied by specialists of the pharmacology field.

One of the examples of usage of AI for identifying suitable drugs in-silico are: Deep learning-based models to predict binding affinities based on chemical sequences (SMILES) and amino acid sequences (FASTA) of a target protein. Drugs like Atazanavir, Remedesivir, Kaletra, Rapamycin and tiotropium bromide were identified as potential inhibitors of the SARS-CoV2 virus (Of these ramdesivir has recently been approved by US FDA).

In addition, many drugs under trial can be critically examined for adverse outcomes using these techniques. An approach known as PrOCTOR has been used to predict side effects of under-trial drugs using Random Forest and Principal component analysis.

The drug discovery landscape, discussed here, shows that repurposed drugs is the fastest way to bring COVID-19 treatment to the general population. In addition to already approved repurposed drugs, there is a need for identifying more repurposed drugs. AI and ML can boost this process by quickly identifying drugs having efficacy against COVID-19 and hence bridge the gap between thousands of repurposed drugs, laboratory /clinical testing and final drug authorisation.

Read the original post:
Artificial intelligence as COVID-19 drug discovery booster - Express Healthcare