Correspondence between representations in visual cortices and neural networks
Artificial mechanism of "attention" through the deep learning
2021-02-08
(Press-News.org) This discovery was made possible by applying the research method for the comparison of the brain activity between monkeys and humans to artificial neural networks. This finding might be helpful not only to understand the cortical mechanism of attentional selection but also to develop artificial intelligence.
Deep neural networks (DNNs), which are used in the development of artificial intelligence, are mathematical models for obtaining appropriate mechanisms to solve specific problems from the training with a large-scale dataset. However, the detailed mechanisms underlying DNNs through this learning process have not yet been clarified.
A research group led by Nobuhiko Wagatsuma, Lecturer at the Faculty of Science, Toho University, Akinori Hidaka, Associate Professor at the Faculty of Science and Engineering, Tokyo Denki University, and Hiroshi Tamura, Associate Professor at the Graduate School of Frontiers Biosciences, Osaka University, found that the characteristics of responses in DNNs for predicting the attention to the most important location in images were consistent with those of the neural representation in the primary visual cortex (V1) of primates. The discovery was made possible by applying the analysis method designed for comparing the characteristics of the neuronal activity in monkeys with that in humans to DNNs.
The result of this study provides important insight into the neural mechanism of attention. Additionally, the application of the attentional mechanism in the primates including human may accelerate the development of artificial intelligence.
Key Points:
The correspondence between primate visual cortices and deep neural networks has been revealed by applying the research method for comparing the neural activity between different species, such as humans and monkeys, to artificial neural networks.
Recently, deep neural networks are utilized as the main methods for developing artificial intelligence. Wagatsuma et al. have reported the similar properties between deep neural networks for predicting attention and the primary visual cortex (V1) of primates. Additionally, their findings implied that the mechanism of the deep neural networks for attention prediction might be distinct from that for object classification such as VGG 16.
Attention is a function that enables us to attend the most important information at the moment, which is the most critical keyword in the recent development of artificial intelligence. The results of this research might provide contributions not only for understanding the neural mechanisms for attention selection of primates including human but also for developing artificial intelligence.
INFORMATION:
These results have been published in eNeuro, an open-access journal of Society for Neuroscience (SfN), headquartered in the United States, on November 24, 2020.
[Attachments] See images for this press release:
ELSE PRESS RELEASES FROM THIS DATE:
2021-02-08
On Feb 5th, Seoul National University, College of Engineering (Dean Kookheon Char) announced that Professor Sang Woo Seo's research team (Dr. Jina Yang and Mr. Yong Hee Han (graduate student)) at School of Chemical and Biological Engineering has developed a synthetic protein quality control system to enhance full-length translation in bacteria. This technology is expected to increase the efficiency of the production of biopharmaceuticals, industrial enzymes, and bio-based chemicals.
Recombinant proteins are used in various industrial fields from protein drugs such as insulin to industrial proteins such as laundry detergents. Since proteins can perform their functions only with full-length and proper 3D structure, recombinant protein production ...
2021-02-08
Against the background of global warming, extreme heat days (EHDs) occur frequently and greatly threaten human health and societal development. Therefore, it is of great importance to understand the variation of EHDs.
Previous studies have indicated that the frequency of EHDs is mainly modulated by the mean state of temperature, and thus the frequency of EHDs mostly presents an increasing trend.
"However, the variability of the daily maximum temperature also plays an important role in the interdecadal change of extreme heat days over Northeast China," says Ms. Liu Wenjun, a Master's student from the group of Dr. Ruidan Chen in the School of Atmospheric Sciences ...
2021-02-08
Researchers from Hong Kong Baptist University (HKBU) and the Faculty of Medicine at The Chinese University of Hong Kong (CU Medicine) have jointly developed the Spermine Risk Score which, coupled with the use of a urine test, provides a non-invasive and more reliable method for the diagnosis of prostate cancer. In a study conducted by the researchers, about 37% of the patients, who were ultimately found to have no prostate cancer, can avoid undergoing a prostate biopsy procedure. The findings have just been published in the scientific journal Prostate Cancer and Prostatic Diseases.
Demand for more reliable and non-invasive diagnosis
Prostate cancer is the third most common and the fourth most fatal cancer for the male ...
2021-02-08
Which bananas end up in your shopping basket-- the uniformly yellow ones or those with brown spots?
If you are like most people, you skip the spotted ones and select those that are perfectly yellow. This is because emotions play an an oversized role in our shopping decisions, according to a new study by Danish and Swedish researchers.
"We choose food based upon an expectation of what it will taste like that is bound to our feelings. So, if we expect a brown banana to not match the taste of a yellow one, we opt for the latter," explains Karin Wendin, an associate professor at University of Copenhagen's Department of Food Science, and one of the researchers behind the study.
Approximately 716,000 tonnes ...
2021-02-08
A team of researchers led by Nanyang Technological University, Singapore (NTU Singapore) has developed a new material, that when electricity is applied to it, can flex and bend forty times more than its competitors, opening the way to better micro machines.
Conversely, when it is bent, it generates electricity very effectively and could be used for better "energy harvesting" - potentially recharging batteries in gadgets just from everyday movements.
The novel material is both electrostrictive and piezoelectric. Its electrostrictive properties means it can change shape when an electric current is applied, while piezoelectric means the material can convert pressure into electric charges.
When an electric field is applied, the atoms that make up electrostrictive ...
2021-02-08
Although SGLT-2 inhibitors are central to the treatment of diabetes, their exact mode of action was hitherto unknown. In a study conducted by a research group led by Peter Wolf, Martin Krssak and Michael Krebs from MedUni Vienna's Department of Medicine III, magnetic resonance spectroscopy (MRS) was used to show that there is a direct correlation between the elimination of glucose via the kidneys and new glucose production in the liver. A single dose of the SGLT-2 inhibitor dapagliflozin gives rise to a beneficial regulation mechanism, in which glucose loss due to drug-induced SGLT-2 inhibition is exactly balanced out by an equal increase in new glucose production in the liver. The study has been published in the leading journal Diabetes Care.
Dapagliflozin is a drug from the group ...
2021-02-08
Researchers at the Kobe University Graduate School of Science have revealed that when captive-bred juvenile red-spotted masu salmon are released into natural streams, very few individuals become migrants.
Red-spotted masu salmon was an important fish species for the fishing industry in the rivers of west Japan, however in recent years their numbers are declining rapidly. The results of this research offer important suggestions for stocking practices and the management of river environments.
The research group consisted of graduate school students TANAKA Tatsuya and UEDA Rui and Associate Professor SATO Takuya. The results were published in Biology Letters ...
2021-02-08
[Background]
The human body consists of about 60 trillion cells that are renewed day by day to maintain homeostasis of body tissues. In particular, cells of the digestive tract are renewed completely within several weeks thanks to vigorous proliferation where tissue stem cells of every tissue play critical roles in supplying those cells. Tissue stem cells play essential roles in various phenomena such as histogenesis and recovery from damage by producing differentiated cells while dividing. They do this by producing identical cells (self-renewal) or by differentiating into other types of cells. The research team led by Profs. Murakami and Barker of the Cancer Research Institute, Kanazawa University revealed ...
2021-02-08
Betelgeuse is normally one of the brightest, most recognizable stars of the winter sky, marking the left shoulder of the constellation Orion. But lately, it has been behaving strangely: an unprecedentedly large drop in its brightness has been observed in early 2020 (Figure 1), which has prompted speculation that Betelgeuse may be about to explode.
To find out more, an international team of scientists, including Ken'ichi Nomoto at the Kavli Institute for the Physics and Mathematics of the Universe (Kavli IPMU), conducted a rigorous examination of Betelgeuse. They concluded that the star is in the early core helium-burning phase (which is more than 100,000 ...
2021-02-08
A study reveals the genetic factors that may expose or protect people with Down syndrome from SARS-CoV-2 infection, as well as the prognosis of COVID-19.
Their findings, which are published in the journal Scientific Reports, follow previous studies showing a tenfold mortality risk of COVID-19 for people with Down syndrome, adding further evidence to boost existing calls for priority vaccination of the medically vulnerable group.
The researchers analysed all publicly available Down syndrome transcriptomic data to uncover alterations that might affect SARS-CoV-2's infection and disease progression.
TMPRSS2, a gene that codes for an enzyme critical for aiding the entry of SARS-CoV-2 in human cells, had 60% higher levels ...
LAST 30 PRESS RELEASES:
[Press-News.org] Correspondence between representations in visual cortices and neural networks
Artificial mechanism of "attention" through the deep learning
