Use Cases of Machine LearningNovember 7, 2017
Will Machines Replace Humans?November 20, 2017
It is important to understand that Machine Learning should not be considered as a “cure all” within your intelligence systems, but merely as a valuable tool to have in your toolbox. There are several challenges that comes with Machine Learning if it is not properly applied and managed, the top three being:
1. Human Judgement Override
There are several tools that can be used to arrive at the fastest time to insight and execution, and relying on only one tool is not recommended. However, it is a challenge to know when and how much human judgement to apply to the model. Sometimes, it is easy to let your preconceived ideas on what you want the analysis to show, or outcome to be, take too much control over the natural course of action.
2. Replacement versus Augmentation
In theory, all Machine Learning should ultimately result in a “single” point of knowledge, i.e. regardless of situation you should arrive at the same outcome as determined by the algorithms. But as that wouldn’t be a very feasible in a real-life scenario, it is very important that Machine Learning reliance consists of a combination of interactions between other algorithms and tools, including human intelligence.
3. Specialized Learning versus Generalized Learning
Some Machine Learning capabilities can be restricted to certain components of the intelligence model, while others can holistically provide industry and global knowledge. Sometimes Machine Learning is used in areas that might not be suited to the design of the Machine Learning algorithm, resulting in contamination of the results that other tools present, thus compromising the overall results of the analysis.
This was part seven in our new blog series, based on the article “Machine Learning Implications for Intelligence and Insights”, written by Jesper Martell, Comintelli, and Paul Santilli, Hewlett Packard Enterprise.