IGLOO Corporation, acquires two artificial intelligence patents
- IGLOO Corporation to help more organizations reap the benefits of AI
[April 12, 2022] Security and intelligence company IGLOO Corporation (CEO Lee Deuk-Choon, www.igloo.co.kr/en) announced that it has completed registration of two artificial intelligence (AI) patents. Through the application of this patent, it is expected that the efficiency of 'labeling' and 'Automated Machine Learning (AutoML)', which are core processes of machine learning (ML), will be improved one step further.
Machine learning, a field of artificial intelligence, is a technology that allows a machine (algorithm) to make its own judgment criteria based on learning from data and make predictions on other data. In order to make good use of the machine learning prediction model modeled through data and algorithms, ‘labeling’ work is required to attach features to various data and transform them into a form suitable for the machine learning prediction model. In addition, in order to quickly select an algorithm that matches the data to be predicted, the 'auto ML' technique that automates repetitive manual tasks is also widely used.
The first patent is a technology that generates training data optimized for web firewalls to which machine learning prediction models are applied, and utilizes prior learning of previously labeled training data. First, the features of the web firewall training data that have been verified and labeled by experts are extracted and learning is performed. After training, the machine learning prediction model classifies unlabeled training data among the data generated by the web firewall and labels it in a form suitable for the web firewall to generate high-quality learning data in a timely manner.
The second patent is a technology that improves the 'auto ML' process so that an algorithm that matches the data to be predicted can be found faster and more accurately. A plurality of algorithms are sequentially applied to the data to be predicted, and the evaluation indicators derived through this are compared and determined from various angles. In the subsequent verification process, the algorithm with the highest accuracy is selected by automatically substituting a plurality of algorithms only for data that yields prediction values different from the labeled values. Through ensemble learning that combines up to two models in this way, it is possible to reduce the time invested in developing the predictive model and increase the accuracy of the prediction result.
Since around 2015, IGLOO Corporation has been operating a team of experts in various fields such as security, big data, data mining, and data visualization, developing algorithms optimized for security and generating high-quality training data. In February 2019, Korea’s first AI security monitoring solution, ‘SPiDER TM AI Edition’, was launched and is being supplied to major public institutions and companies.
Lee Deuk-Choon, CEO of IGLOO Corporation, said, “For many years, IGLOO Corporation has been conducting AI research to develop machine learning prediction models and increase the efficiency of generating training data. We will continue to strengthen AI R&D and support more organizations to benefit from AI.”