Intelligence Processor: An unstructured data processing machine

Current processor is designed for processing structured data using pre-defined paradigm, i.e. the input, output, and storage are accurate and the processing rules are determined. However, most of natural data are unstructured, such as video, image, and voice. Although many efforts have taken to process unstructured data using predefined algorithm or a neural network, they still have weak expression ability for complicated reasoning tasks which may have many steps and complicated logical relationships. In such cases, the rules and experiences are applied as a ‘program’ to process the data. Differs from the current processor’s program, these programs called ‘experience rules’ are inaccurate and are obtained from unstructured data previously, or are taught following a teaching mechanism. In this work, a newly established processor architecture is introduced using the experience rules as the program. The processor is designed for obtaining program from unstructured data following basic learning rules. Then the experience rules are applied for processing new unstructured data. The processor is designed for animating the thinking procedure of human beings, which paves the way to the basic principles of the artificial general intelligence (AGI) processing machine.

 

Fig.1. Comparison between proposed intelligence processor and the traditional processor.

 

 

 

 

  

  • 1
    点赞
  • 0
    收藏
    觉得还不错? 一键收藏
  • 0
    评论
This book explores the significant role of granular computing in advancing machine learning towards in-depth processing of big data. It begins by introducing the main characteristics of big data, i.e., the five Vs—Volume, Velocity, Variety, Veracity and Variability. The book explores granular computing as a response to the fact that learning tasks have become increasingly more complex due to the vast and rapid increase in the size of data, and that traditional machine learning has proven too shallow to adequately deal with big data. Some popular types of traditional machine learning are presented in terms of their key features and limitations in the context of big data. Further, the book discusses why granular-computing-based machine learning is called for, and demonstrates how granular computing concepts can be used in different ways to advance machine learning for big data processing. Several case studies involving big data are presented by using biomedical data and sentiment data, in order to show the advances in big data processing through the shift from traditional machine learning to granular-computing-based machine learning. Finally, the book stresses the theoretical significance, practical importance, methodological impact and philosophical aspects of granular-computing-based machine learning, and suggests several further directions for advancing machine learning to fit the needs of modern industries. This book is aimed at PhD students, postdoctoral researchers and academics who are actively involved in fundamental research on machine learning or applied research on data mining and knowledge discovery, sentiment analysis, pattern recognition, image processing, computer vision and big data analytics. It will also benefit a broader audience of researchers and practitioners who are actively engaged in the research and development of intelligent systems.
The ideas introduced in this book explore the relationships among big data, machine learning and granular computing. In many studies, machine learning has been considered as a powerful tool of big data processing. The relationship between big data and machine learning is very similar to the relationship between resources and human learning. In this context, people can learn from resources to deal with new matters. Similarly, machines can learn from big data to resolve new problems. However, due to the vast and rapid increase in the size of data, learning tasks have become increasingly more complex. In this context, traditional machine learning has been too shallow to deal with big data sufficiently, so granular computing concepts are used in this book to advance machine learning towards the shift from shallow learning to deep learning (in its broader sense). The focus of this book is on the development and evaluation of granular computing based machine learning approaches in terms of classification accuracy. In this context, the authors consider traditional machine learning to be of single-granularity and the proposal of granular computing based machine learning is aimed at turning single-granularity learning into multi-granularity learning. In particular, the authors proposed the following transformations: (a) supervised learning to semi-supervised learning, (b) heuristic learning to semi-heuristic learning, (c) single-task learning to multi-task learning, (d) discriminative learning to generative learning and (e) random data partitioning to semi-random data partitioning. In addition, the authors also explore how to achieve in-depth evaluation of attribute-value pairs towards induction of high-quality rules, in the setting of multi-granularity learning.

“相关推荐”对你有帮助么?

  • 非常没帮助
  • 没帮助
  • 一般
  • 有帮助
  • 非常有帮助
提交
评论
添加红包

请填写红包祝福语或标题

红包个数最小为10个

红包金额最低5元

当前余额3.43前往充值 >
需支付:10.00
成就一亿技术人!
领取后你会自动成为博主和红包主的粉丝 规则
hope_wisdom
发出的红包
实付
使用余额支付
点击重新获取
扫码支付
钱包余额 0

抵扣说明:

1.余额是钱包充值的虚拟货币,按照1:1的比例进行支付金额的抵扣。
2.余额无法直接购买下载,可以购买VIP、付费专栏及课程。

余额充值