The next generation of artificial intelligence (AI) used in public health, IoT, and other mission applications will soon be able to make better decisions and more accurate predictions with some of the philosophical wisdom. Georgia was founded in Tech.
Researchers from the School of Computer Science and Engineering (CSE) receive a $1.1 million grant from the National Science Foundation (NSF) over an average of 3 years to use time series AI models to measure uncertainty in current AI models. Can be found, one can guess.
From there, they hope to teach a model known as the Socratic paradox: “I know that I know nothing.”
The amount of uncertainty is allowed to say ‘I don’t know the uncertainty model’ when faced with an unknown or unexpected situation. B Aditya Prakash, Associate Professor CSE and Principal Investigator of the project, said.
Because it doesn’t know what it doesn’t know, the current model can guess the answers and proceed as if it were predicted correctly. This is very problematic with time-series data – such as public health surveillance and forecasting – where inaccurate estimates and answers can lead to low confidence in predictions produced by current-generation models.
Prakash and Chao Zhang , assistant professors and principals of CSE, work with co-pei Shokoti Yao George Mason University to remove these limitations. In addition to measuring the amount of uncertainty, the team works to better understand the types and sources of future uncertainty.
Because many depend on certain functions and data, the researchers say it is difficult to provide accurate estimates for increasing efficiency based on their new approach. However, initial disease prediction results suggest that the model, algorithm, and new technology team can perform a principal component model with up to 2.5 times accuracy and up to 2.4 times reliability.
The technology and tools resulting from this project, officially Collaborative Research: NSF Award #2106961, will be the original quantification of what is not appropriate, open-source, open-source, and courses, tutorials, and more Workshop More workshops integrate research results.
Facebook Search Award:
Prakash and Zhang took other approaches to improve the existing modeling. The new pairing won the Facebook 2021 Statistics for Insights, Models, and Better Decisions awards for FALSE. Their proposal is one of 10 winners announced by new Facebook research because it predicts time series calibrated by hierarchy.
According to the team, the non-parametric statistical method can be very flexible and efficient for modeling time series data when there are unknowns in many data sets. This is because non-acting instruments analyze group media rather than group media. As a result, scientific openings can be better understood and used to strengthen them.
We will also integrate signals from different ideas, such as demographic signals, time-series signals, and mechanical models, of where they interact with each other. The model can equally strengthen to make it more accurate, reliable, and robust,” Zhang said.
After the project is completed, the study’s results will be used in various industrial applications to improve health care, public health, and future modeling.
For example, Prakash said, Facebook could use this technology to predict demands in data centers in different geographic regions. This will give them a big-time to make their infrastructure stronger and more efficient.
Founded in 1955, M&M from the end of the sixth century, and until M, actually Fronicist from the sixth century AD. Fondi provides readers with articles and book records from top scholars in Europe and North America. The publishing language is English, although papers are also published in Latin, French, German, and Italian.
Whats a socratic paradox?
It is a paradox, but an informal paradox, and can be “resolved”, or at least discussed, because it is not Aristotle’s argument between the different meanings of the word “address”. , for example, one can distinguish between reflective knowledge (knowledge of my knowledge) and nonconverging knowledge (to know about some facts). There is no hesitation in saying that I know that I do not have knowledge that does not think. The first type of knowledge is not the same as the second, as they are qualitatively different.
It is important to distinguish between formal contrasts and informal contrasts. Informal paradoxes are not annoying because informal logic is tolerant. Or it can make a subtle difference on an ad-hoc basis. The danger is not because contradictions will destroy the logical fabric of the universe but because sophistication will abuse it. Formal logic, in contrast, is abstract and intolerable, as it does not allow judgments of judgment to consider whether utterances are well-formed that do not appear in the rules governing formalisms.