The hottest IBM shows the research progress of art

  • Detail

IBM shows the research progress of artificial intelligence in 2018 and predicts the future trend

recently, IBM showed its progress in the field of artificial intelligence (AI) in 2018 and predicted three main trends of AI development in the future. IBM selected 100 integrated papers from the AI papers written by researchers and scientists in 12 laboratories around the world, and showed many progress made in AI technology development, technology application and enhancing trust in AI

in speech recognition, the AI system developed by IBM can not only answer simple questions, but also understand the most common arguments among interlocutors; In terms of image recognition, a new learning method is developed, which no longer needs to learn thousands of marked images. Only the accuracy of standard dynamometer can reach about 0.15%, and a small number of examples are needed to accurately recognize images; Through an innovative architecture and algorithm to realize machine interactive learning, the two AI systems can teach and cooperate with each other, which further improves the learning efficiency of AI; In answering developmental questions, a new method is adopted, which can rearrange and organize materials across paragraphs to generate more accurate answers. Many advances have also been made in eliminating AI bias and improving the transparency of the algorithm, including breaking the "black box" of the algorithm and enhancing the ability of neural networks to resist attacks. The progress involving AI technology itself includes using new methods to improve the efficiency of deep neural network learning and training, accelerate the speed of deep neural network reasoning, and reduce the workload of designing neural networks

on the basis of reviewing the progress, IBM also predicted three trends of AI development in the future

one is the transformation of AI from correlation analysis to causal reasoning. In 2019, causal expectation modeling technology will become the core of the development of artificial intelligence through the change of electromechanical speed energy. Traditional AI is basically based on correlation analysis and lacks a deep understanding of causality. The new causal reasoning method can infer the causal relationship from the data and test it effectively, so as to make better decisions

second, develop seven tips for trusted people to choose universal tensile machine. Industrial intelligence will become the focus of research. In order to deal with the impact of AI on ethics, many institutions have established ethics advisory committees to deal with data leakage, consumer privacy protection and other issues; In terms of research, the fairness, interpretability and robustness of the algorithm are increased, which helps to increase the R & D investment in AI trust; In terms of application, it is emphasized to deploy more artificial intelligence that will help improve social welfare. In 2019, these efforts will become the core of enterprises' construction, training and deployment of AI technology

the third is the combination of quantum technology and AI. The development of quantum computing will provide assistance for AI. Quantum computing provides new possibilities to improve the speed and efficiency of computing and meet the requirements of large-scale computing. At present, thousands of enterprises have accessed IBM's quantum cloud computing services. With the increasing complexity of AI problems, quantum computing may change the way we deal with AI computing tasks

Copyright © 2011 JIN SHI