Despite the glamor associated with the term ‘machine learning’, the reality is often tedious and repetitive in nature. This makes many forms of machine learning ideal for the application of artificial intelligence. The performance measure is fairly straightforward (better accuracy), and the variables are quite easy to define – parameters for algorithms, choice of algorithms, features, data sets – and so on. As a result AI can dramatically improve the efficiency of machine learning projects, and relieve the data scientist of substantial tedium. Several platforms that provide AI assisted machine learning are listed below – and more will be added in coming months as these new platforms come on stream.
CrowdFlower is focused on making data useful by helping data teams collect, clean and label their data at scale. The CrowdFlower platform combines human-labelled training data, easy to deploy machine learning, and human-in-the-loop capabilities to help companies create value from their unstructured data.
DataRobot is a cloud based machine learning service that takes much of the grunt out of predictive model building. It automatically searches for the best features, selects the most appropriate algorithms, tests the model and provides an API for model deployment. It takes best-in-class algorithms from R, Python, H20, Spark, and other sources, and employs text mining, variable type detection, encoding, imputation, scaling, transformation, and automated feature engineering. The infrastructure is massively powerful for rapid processing, and big data is well supported with certification on Cloudera Enterprise 5. A free community version is available at the time of writing.
Fuzzy.ai is an API that makes it easy for developers to build decision-making artificial intelligence without needing data science expertise or lots of data. Fuzzy.ai’s team have previously been involved with designing and building Gmail, Google Calendar, Drupal, WikiTravel and Breather. Fuzzy.ai is used for price optimization, recommendations, lead scoring and matching in 2-sided marketplaces.
Pavlov uses AI to make data science accessible via a web service.
PurePredictive uses AI to automate the machine learning process. The platform automates the discovery of data transformations and higher order relationships between data features and automatically accommodates data drift. The cloud platform scales automatically for workloads so that data sets of virtually any size can be accommodated. The models are easily consumed through web services, and can be automatically maintained to deal with changes in business conditions.
Scaled Inference – will be offering machine learning as a service, using AI to guide users through the process.
Seldon is an open source and platform-agnostic machine learning platform that helps data scientists build intelligence aided by AI.