Blog: Artificial Intelligence and Machine Learning: Advances, Limitations, and Trends
Artificial intelligence and machine learning particularly form the foundation of many successful disruptive industries we are currently witnessing. These industries include, but are not limited to, autonomous driving, cognitive healthcare, assistive technologies, precision agriculture, home intelligence and Industry 4.0 to name just a few. Machine learning is a sub-field of artificial intelligence that endows an artificial system/process with the ability to learn from experience and from observation without being explicitly programmed. Traditional machine learning techniques follow top-down deductive reasoning approach where they start out by formulating abstract and wide-ranging hypotheses about the world from preexisting knowledge and few examples. They make predictions about what the data should look like if those hypotheses are correct and then revise their hypotheses, depending on the outcome of those predictions and upon receiving more observations. In contrast, statistical machine-learning algorithms are characterized as bottom-up inductive reasoning (working on numeric data to infer symbols or inferring general rules from a set of examples) and try to extract patterns from raw data (thousands or millions of examples). These modern statistical machine-learning techniques have powered different products and services through endowing them with the ability to discover and recognize complex patterns and trends (class, cluster or anomalies) in archived, streaming or live data and predicting future values and to learn from examples and from observations at different levels of abstractions. This talk provides a comprehensive introduction to the key artificial intelligence and machine learning concepts and techniques involved in the development of intelligent products and services. It describes some limitations of the current weak/narrow artificial intelligence and statistical machine learning techniques and sheds light on the recent advances in machine learning to handle these limitations.
Keynote speech at The second International Conference of Computing and Informatics 2019 (ICCI-2019)