Machine Learning

Created by: Felix Weber

Machine Learning is a significant element of AI and is so popular that is sometimes confused with Artificial Intelligence. In spite of this huge emphasis on machine learning, the key driver for the growth of AI is actually Deep Natural Networks (DNNs). These networks display the ability of learning patterns of video, image and speech in a much more automatic and faster manner. They enhance the capabilities of ML and have altered the way an ML engineer performs.

Table of Content [Hide]

Machine Learning is a significant element of AI and is so popular that is sometimes confused with Artificial Intelligence. In spite of this huge emphasis on machine learning, the key driver for the growth of AI is actually Deep Natural Networks (DNNs). These networks display the ability of learning patterns of video, image and speech in a much more automatic and faster manner. They enhance the capabilities of ML and have altered the way an ML engineer performs [1].

Machine Learning aims at devising new ways of building strong predictions automatically from complicated data. It has a close association with advanced Statistics and most of the ideas associated with ML have been the brain work of Statisticians. The key focus of statisticians has been on model inference but the ML community focuses on maximization of predictive performance [2]. The ML field is standardized against various out-of-sample illustrations that assess the extent to which new data is predicted on a newly trained model. Efforts are being made to enhance the transparency of Machine Learning but on the other hand ML practitioners do not wish to assign structural meanings to certain models.

Such models are evident as black boxes aimed at predicting the future based on patterns similar to the ones used in the past. Machine learning can be defined as a technology which permits a system to learn from experiences, data and examples. Artificial Intelligence on one hand is the science of transforming machines into smart ones and machine learning on the other hand is the technology which assists computers in performing particular tasks intellectually. Such systems are capable of implementing complicated processes as they learn from existing data and do not follow pre-programmed rules blindly. There have been significant developments in the abilities of machine learning due to technical advancements; increase in computing power and better availability of all data types. These advancements have helped systems to outperform individuals in certain tasks. There are a number of object identification and exist voice systems that outperform humans even though they are limited in nature [3]. Today, humans interact in their daily lives with ML driven systems for image recognition used for tagging pictures on social media; they are used for voice identification used mainly by virtual personal assistants and is also used in recommender systems used by online retailers.

Apart from the existing applications, it seems that the field has great potential as a number of ML applications are under development in different fields like transport, education, healthcare or retail and wholesaling[4]. ML is capable of providing precise personalized diagnosis or health diagnostics; customizing classroom activities for advanced learning and in supporting transport systems. It also supports scientific developments by acquiring insights from huge datasets and triggers efficient operations in various industrial sectors. Advanced ability to learn from insights of big data sets assists machine learning in increasing the productivity; offer efficient public services and create advanced goods or services customized to individual requirements [3]. Nevertheless, queries regarding advanced data usage and the role played by intellectual computer systems in modern society still remain unanswered.

This technology offers innumerable benefits and its pervasiveness is increasing with passing time. Thus, it is high time that it is further developed such that it stimulates public confidence and deals with the key challenges or concerns. This will not only help in addressing the key ML risks but also make sure that the different benefits are realized.

References

  1. Yoshua Bengio; Learning Deep Architectures for AI. Foundations and Trends® in Machine Learning 2009, 2, 1-127, 10.1561/2200000006.
  2. Fabrizio Sebastiani; Machine learning in automated text categorization. ACM Computing Surveys 2002, 34, 1-47, 10.1145/505282.505283.
  3. Christian Robert; Machine Learning, a Probabilistic Perspective. CHANCE 2014, 27, 62-63, 10.1080/09332480.2014.914768.
  4. Felix Weber; Reinhard Schütte; A Domain-Oriented Analysis of the Impact of Machine Learning—The Case of Retailing. Big Data and Cognitive Computing 2019, 3, 11, 10.3390/bdcc3010011.