The nervous system and large data, the new dimensionality reduction algorithm makes the brain easier

Scientists have now been able to detect and record the activity of hundreds of neurons in the brain and continue to develop technological commitments to multiply this number. However, simply recording neuronal activity does not allow scientists to automatically understand how the brain works.

In a commentary published in the journal Nature Neuroscience, Byron M. Yu, of Carnegie Mellon University, and John P. Cunningham of Columbia University described many scientific motivations for studying the activities of a large number of neurons, to explain the activity of neurons, A machine learning algorithm named dimensionality reduction is proposed.

In recent years, dimensionality reduction has given us an insight into how the brain distinguishes between smells, how to make decisions in the face of uncertainty, and how to think about moving limbs when there is no real action. Yu and Cunningham Advocate dimensionality reduction as a standard analytical method that will make it easier to compare healthy and abnormal brain activity patterns and ultimately improve treatment and intervention for brain damage and brain dysfunction.

"One of the core principles of neuroscience is that a large number of neurons work together to produce brain function," said Yu, an assistant professor and CNBC staff at CMU Electrical and computer Engineering and biomedical engineering. However, the most standard analysis method can only analyze one or two neurons at a time. To understand how large numbers of neurons interact, advanced statistical methods such as dimensionality reduction can explain these massive neurological records. ”

The real idea of dimensionality reduction is to use less latent or hidden variables to sum up a lot of neuronal activity. The method of dimensionality reduction is extremely useful in uncovering the internal working mechanisms of the brain, for example, when we meditate or solve mental and mathematical problems, all of which are carried out in the brain, not in the external world. These latent variables can be used to depict the thinking path.

"The main goal of scientific research is to explain complex phenomena in simple terms," Cunningham, an assistant professor of statistics in Cu, said in the article. Traditional neuroscientists aim to find a way to simplify a single neuron. But they are becoming increasingly aware that the variety of features that neurons exhibit in active patterns is difficult to explain by examining a neuron. dimensionality reduction provides a way to cover the heterogeneity of individual neurons and to find a simple explanation based on how neurons interact with each other. ”

Although in neuroscience, dimensionality reduction is relatively new compared to existing analytical methods, it has shown great potential and bright prospects. With the continuous development of neural recording technology and the Developing of the "Brain Program (BRAIN Initiative)" In the United States, large data is becoming more and more important, and the use of the method of reducing and keeping the peace will become an essential data processing method.

Related reading: Discovery of the new pathways controlling the serotonergic system

Original link: Neuroscience and big data:how to find simplicity in the brain (translation/Shaohong leaf Zebian/Zhonghao)

Free Subscription "CSDN cloud Computing (left) and csdn large data (right)" micro-letter public number, real-time grasp of first-hand cloud news, to understand the latest big data progress!

CSDN publishes related cloud computing information, such as virtualization, Docker, OpenStack, Cloudstack, and data centers, sharing Hadoop, Spark, Nosql/newsql, HBase, Impala, memory calculations, stream computing, Machine learning and intelligent algorithms and other related large data views, providing cloud computing and large data technology, platform, practice and industry information services.