Custom Cognitive Computing Aims To Turn Every Developer Into A Data Scientist


Today, AI can be consumed in two forms – 1) cognitive APIs, and 2) custom ML models. Cognitive APIs offer computer vision, speech, translation, text analytics as cloud-hosted APIs. IBM Watson, Google Cloud ML APIs, Microsoft Cognitive Services and Amazon AI APIs are examples of such hosted APIs. Any developer with the basic understanding of consuming REST API can invoke these services to add intelligence to her application. But these services come with a limitation – they are trained on generic datasets. That means while computer vision API can identify a car as an object, it cannot detect its make and model. If the developer is building an application that has to precisely determine the car make and model, these APIs won't work.