FEATURE : MACHINE LEARNING early development stage , the region is seeing it revolutionising a range of industries , with research and development advances being made in Machine Learning every day . “ For enterprises to ensure the success and ROI of Machine Learning deployments , it is important for them to align them to defined clear goals and use cases , and associate these to business priorities . “ Identifying and understanding whether the problems they are trying to solve could be tackled better and more accurately by Machine Learning rather than conventional software is key . Additionally , having experts run an elaborate experimentation phase of the potential projects which includes everything from gathering and assessing data , to basic modelling , cost and risk assessment can help predict whether the project will be successful or not ,” he added . “ This requires nurturing an organisation culture that values innovation . In the near future , we can expect quantum computing to significantly increase the capabilities of Machine Learning . It will give Machine Learning the capability to create systems that execute multi-state operations simultaneously . Quantum Machine Learning will have the ability to tackle complex issues in a split second .”
According to Jacobson , many organisations face challenges in moving Machine Learning models into production environments . On average , between 60 % and 80 % of models created with the intent to deploy are never deployed . “ Plus , it typically takes six to eight months to deploy a model using legacy technologies , which leads to many projects becoming obsolete before they can go live ,” he noted . “ Machine Learning models are not a ‘ one and done ’ exercise . Model management , which can involve monitoring , revisiting and retraining , is a fundamental part of their life cycle . How you put your model into production will determine how easy it is to manage . Yet , even getting into production could be challenging .”
In fact , said Jacobson , this step was the highest hurdle cited in Davenport ’ s Machine Learning research , with 47 % of executives saying that it has been difficult to integrate Machine Learning projects with existing processes and systems .
“ Enterprises that struggle to integrate Machine Learning applications with existing production applications waste time and money on data science projects that are never put into production ,” he said . ” Machine Learning operations ( MLOps ) is the critical process that makes this possible by treating ML and other types of models as reusable software artifacts . Models can then be deployed and continuously monitored and retrained via a repeatable process . As such , it helps businesses discover valuable information and insights from their data more quickly .”
Heriot-Watt University ’ s Gill said one of the most common Machine Learning challenges that enterprises face is the availability of data . “ Proper access to raw data is very important as training Machine Learning algorithms requires huge amounts of data ,” he said . “ Insufficient data makes it harder to train the models properly which makes the implementation of Machine Learning project even harder .”
Gill said in addition to proper collection of data , enterprises also need to model and process the data to fit the algorithms that they will be using . “ Another major challenge faced with Machine Learning implementations is data security . Distinguishing between sensitive and insensitive data is crucial for implementing Machine Learning correctly and efficiently ,” he said . “ Sensitive data should be encrypted and stored in other servers or at a location where the data is fully secured . Only trusted team members should be allowed access to confidential data . p
www . intelligentcio . com INTELLIGENTCIO AFRICA 37