Intel unveils its first AI-driven neural network chip

Agencies
August 21, 2019

San Francisco, Aug 21: Intel has unveiled its first high-performance Artificial Intelligence (AI)-driven neural network chip "Nervana" to meet the growing high-speed computing and robotics demand.

The chip-maker showcased the Nervana neural network processors, with the NNP-T for training and the NNP-I for inference at the "Hot Chips 2019" conference at Stanford University in California on Tuesday.

"To get to a future state of 'AI everywhere,' we'll need to address the crush of data being generated and ensure enterprises are empowered to make efficient use of their data, processing it where it's collected when it makes sense and making smarter use of their upstream resources," said Naveen Rao, Intel Vice President and General Manager, Artificial Intelligence Products Group.

Built from the ground up to train deep learning models at scale, Intel Nervana NNP-T (Neural Network Processor) pushes the boundaries of deep learning training.

It is built to prioritise two key real-world considerations: training a network as fast as possible and doing it within a given power budget.

This deep learning training processor is built with flexibility in mind, striking a balance among computing, communication and memory, said Intel.

Intel Nervana NNP-I, or Springhill, is purpose-built for inference and is designed to accelerate deep learning deployment at scale, introducing specialised deep learning acceleration while leveraging Intel's 10nm process technology with Ice Lake cores to offer industry-leading performance.

Additionally, the Intel Nervana NNP-I offers a high degree of programmability without compromising performance or power efficiency.

"Data centres and the cloud need to have access to scalable general purpose computing and specialized acceleration for complex AI applications," said Rao.

Comments

Add new comment

  • Coastaldigest.com reserves the right to delete or block any comments.
  • Coastaldigset.com is not responsible for its readers’ comments.
  • Comments that are abusive, incendiary or irrelevant are strictly prohibited.
  • Please use a genuine email ID and provide your name to avoid reject.
Agencies
July 2,2020

The American pharmaceutical giant Pfizer Inc. and the European biotechnology company BioNTech SE have conducted an experimental trial of a COVID-19 vaccine candidate and found it to be safe, well-tolerated, and capable of generating antibodies in the patients.

The study, which is yet to be peer-reviewed, describes the preliminary clinical data for the candidate vaccine -- nucleoside-modified messenger RNA (modRNA), BNT162b1.

It said the amount of antibodies produced in participants after they received two shots of the vaccine candidate was greater than that reported in patients receiving convalescent plasma from recovered COVID-19 patients.

"I was glad to see Pfizer put up their phase 1 trial data today. Virus neutralizing antibody titers achieved after two doses are greater than convalescent antibody titers," tweeted Peter Hotez, a vaccine scientist from Baylor College of Medicine in the US, who was unrelated to the study.

Researchers, including those from New York University in the US, who were involved in the study, said the candidate vaccine enables human cells to produce an optimised version of the receptor binding domain (RBD) antigen -- a part of the spike (S) protein of SARS-CoV-2 which it uses to gain entry into human cells.

"Robust immunogenicity was observed after vaccination with BNT162b1," the scientists noted in the study.

They said the program is evaluating at least four experimental vaccines, each of which represents a unique combination of mRNA format and target component of the novel coronavirus, SARS-CoV-2.

Based on the study's findings, they said BNT162b1 could be administered in a quantity that was well tolerated, potentially generating a dose dependent production of immune system molecules in the patients.

The research noted that patients treated with the vaccine candidate produced nearly 1.8 to 2.8 fold greater levels of RBD-binding antibodies that could neutralise SARS-CoV-2.

"We are encouraged by the clinical data of BNT162b1, one of four mRNA constructs we are evaluating clinically, and for which we have positive, preliminary, topline findings," said Kathrin U. Jansen, study co-author and Senior Vice President and Head of Vaccine Research & Development, Pfizer.

"We look forward to publishing our clinical data in a peer-reviewed journal as quickly as possible," Jansen said.

According to Ugur Sahin, CEO and Co-founder of BioNTech, and another co-author of the study, the preliminary data are encouraging as they provide an initial signal that BNT162b1 is able to produce neutralising antibody responses in humans.

He said the immune response observed in the patients treated with the experimental vaccine are at, or above, the levels observed from convalescent sera, adding that it does so at "relatively low dose levels."

"We look forward to providing further data updates on BNT162b1," Sahin said.

According to a statement from Pfizer, the initial part of the study included 45 healthy adults 18 to 55 years of age.

It said the priliminary data for BNT162b1 was evaluated in 24 subjects who received two injections of 10 microgrammes ( g) and 30 g -- 12 subjects who received a single injection of 100 g, and 9 subjects who received two doses of a dummy vaccine.

The study noted that participants received two doses, 21 days apart, of placebo, 10 g or 30 g of BNT162b1, or received a single dose of 100 g of the vaccine candidate.

According to the scientists, the highest neutralising concentrations of antibodies were observed seven days after the second dose of 10 g, or 30 g on day 28 after vaccination.

They said the neutralising concentrations were 1.8- and 2.8-times that observed in a panel of 38 blood samples from people who had contracted the virus.

In all 24 subjects who received two vaccinations at 10 g and 30 g dose levels, elevation of RBD-binding antibody concentrations was observed after the second injection, the study noted.

It said these concentrations are 8- and 46.3-times the concentration seen in a panel of 38 blood samples from those infected with the novel coronavirus.

At the 10 g or 30 g dose levels, the scientists said adverse reactions, including low grade fever, were more common after the second dose than the first dose.

According to Pfizer, local reactions and systemic events after injection with 10 g and 30 g of BNT162b1 were "dose-dependent, generally mild to moderate, and transient."

It said the most commonly reported local reaction was injection site pain, which was mild to moderate, except in one of 12 subjects who received a 100 g dose, which was severe.

The study noted that there was no serious adverse events reported by the patients.

Citing the limitations of the research, the scientists said the immunity generated in the participants in the form of the T cells and B cells of their immune system, and the level of immunity needed to protect one from COVID-19 are unknown.

With these preliminary data, along with additional data being generated, Pfizer noted in the statement that the two companies will determine a dose level, and select among multiple vaccine candidates to seek to progress to a large, global safety and efficacy trial, which may involve up to 30,000 healthy participants if regulatory approval to proceed is received.

Comments

Add new comment

  • Coastaldigest.com reserves the right to delete or block any comments.
  • Coastaldigset.com is not responsible for its readers’ comments.
  • Comments that are abusive, incendiary or irrelevant are strictly prohibited.
  • Please use a genuine email ID and provide your name to avoid reject.
Agencies
April 15,2020

Dear parents, if you want your children to have proper sleep, read this carefully. Joining a growing list of studies that tell parents to shun devices at bed-time, researchers say that children who use devices and decide what time they go to sleep, achieve less sleep and feel more sleepier the following day than their peers.

The study of children in this age-group (aged 11 to 13 years), published in the New Zealand Medical Journal, found most (72 per cent) of the 163 students interviewed by University of Otago researchers achieved recommended guidelines of an average 9 to 11 hours sleep nightly over one week.

"But that also means that almost one in four students did not achieve sleep within these guidelines, which highlights an area for improvement," said study researcher Kate Ford.

However, consistent with previous research in 15 to 17-year-old New Zealanders, the study results show less sleep on the nights where devices are used in the hour before bed.

According to the researchers, students who used devices before going to sleep were also more likely to report that they felt sleepy the following morning. Watching television before bed had no significant effect on sleep length.

There were also some interesting observations over the weekends where students went to bed later but woke later achieving similar sleep length to the school days, the researchers said.

A small group of students (six per cent) who reported less than seven hours of sleep, including a small number reporting not sleeping at all, according to the study,

Therefore, while the average across the week of 72 per cent of students reporting adequate sleep is reassuring, it is far from the goal of every child achieving sleep within the recommended guidelines," Ford said.

Dr Paul Kelly, head of the Sleep Health Service at Canterbury District Health Board, supervised the study and explained that the foundations for good health are based on proper nutrition, regular exercise and good sleep quality.

Sleep quality is often overlooked as a contributory factor to poor health.

"The study findings suggest the need for parental guidance around bedtimings and moderation of the use and availability of electronic devices before bed," Kelly said.

"Respect and protect your sleep, as good daytime functioning is reliant on adequate sleep," Kelly added.

Comments

Add new comment

  • Coastaldigest.com reserves the right to delete or block any comments.
  • Coastaldigset.com is not responsible for its readers’ comments.
  • Comments that are abusive, incendiary or irrelevant are strictly prohibited.
  • Please use a genuine email ID and provide your name to avoid reject.
Agencies
June 23,2020

The record levels of new daily COVID-19 cases are due to the fact that the pandemic is peaking in a number of big countries at the same time and reflect a change in the virus' global activity, the World Health Organisation said.

At a media briefing on Monday, WHO's emergencies chief Dr Michael Ryan said that the numbers are increasing because the epidemic is developing in a number of populous countries at the same time.

Some countries have attributed their increased caseload to more testing, including India and the US But Ryan dismissed that explanation.

We do not believe this is a testing phenomenon, he said, noting that numerous countries have also noted marked increases in hospital admissions and deaths neither of which cannot be explained by increased testing.

There definitely is a shift in that the virus is now very well established, Ryan said. The epidemic is now peaking or moving towards a peak in a number of large countries.

He added the situation was definitely accelerating in a number of countries, including the US and others in South Asia, the Middle East and Africa.

Comments

Add new comment

  • Coastaldigest.com reserves the right to delete or block any comments.
  • Coastaldigset.com is not responsible for its readers’ comments.
  • Comments that are abusive, incendiary or irrelevant are strictly prohibited.
  • Please use a genuine email ID and provide your name to avoid reject.