The artificial intelligence education image data set developed by the distributed AI solution providers was a significant success in the Google platform KAGGLE.
The list of various tools of OORT’s Kagle Data Set was released in early April. Since then, I have climbed to the first page of several categories. KAGLE is a Google -owned online platform for data science and machine learning competition, learning and collaboration.
Ramkumar SUBRAMANIAM, a key contributor to Crypto AI Project OpenLEDGER, said to Cointelewraph, “The attribution page is a powerful social signal, and the data set is to the data scientist, machine learning engineers and practitioners. It indicates that you are participating.
Max Li, the founder and CEO of OORT, observed a promising metric that verifies the initial demand and relevance of educational data collected through the distributed model in COINTELEGRAPH. He added:
“The organic interest in the community, including aggressive use and donations, shows how the distributed community -based data pipelines, such as OORT, can achieve fast distribution and participation without relying on central intermediaries.”
LI also said that OORT plans to launch multiple data sets in the next few months. Among them, the in -vehicle voice command data set, one for the Smart Home Voice command and the other is for the DeepFake video to improve AI -based media verification.
relevant: AI agents are coming to Defi. Wallet is the weakest link.
The first page of several categories
Independently confirmed that the data set reached the first page of KAGLE’s general AI, sleeve and shopping, manufacturing and engineering category by Cointelegraph. At the time of publication, I lost its position on May 6 and on May 14.
Subramaniam recognized the achievements and told Cointelegraph, “It’s not a decisive indicator of actual adoption or enterprise quality.” He said that it is not only “ranking but also the source and incentive class of the data set,” he said. He explained:
“Unlike centralized suppliers that can rely on opaque pipelines, transparent token incentive systems provide a continuous improvement that can be assumed that there is a trace, community cue and correct governance.”
Rex Sokolin, a partner of the AI venture capital company’s venture venture, thinks these results are difficult to duplicate, but “I can organize economically valuable activities using decentralized incentives.”
relevant: Sweat wallet adds AI assistant and expands to multi -chicken dipper.
High quality AI training data: lack of products
The data released by the AI researcher EPOCH AI estimates that the text AI education data created by humans will be exhausted in 2028. Pressure is enough enough to mediate a deal with copyrighted data for AI companies.
Report on how to increase AI education data and how to limit the growth of space has been circulating for many years. Synthesis (AI creation) data is being used more and more at least to some degree of success, but human data is still considered a better alternative that leads to a better AI model.
Contrary to the image of AI training, the situation is becoming more and more complicated with artists who intentionally interfere with training efforts. Hightshade allows users to “poison” and seriously reduce model performance to prevent use of images for AI education without permission.
SubramaManiam said, “We are in an age when high quality image data becomes increasingly lacking.” He also recognized that this tribe is more serious as the popularity of image addiction increases.
“Open source data set is faced with double challenges due to the increase in technologies such as image clocking and hostile watermarks to poison AI training: quantity and trust.”
In this situation, SUBRAMANIAM said that it is more valuable at any time. According to him, such a project said, “It can be a pillar of AI alignment and source in the data economy as well as alternatives.
magazine: AI EYE: AI’s training AI content Go MAD, is the thread loss leader of AI data?