What Important Issues Will Machine Learning Solve in the Next Few Years?

What important world issues do you envision machine learning solving in the next few years? originally appeared on Quora - the knowledge sharing network where compelling questions are answered by people with unique insights.

Answer by David Barber, machine learning researcher: AI, bayesian reasoning, deep learning, on Quora.

Personally I view machine learning as `data driven AI', in particular the attempt to replicate aspects of human perceptual and cognitive abilities. The ML approach is based on measurable performance on real problems. I'm not sure about which world problems might be solved by such an approach.

Certainly the tools that ML and related fields have produced recently have made great progress in for example visual object recognition, speech recognition and language understanding. I don't know if these are `world problems' as such, but I'd like to think that machine learning will help with the organisation and search of knowledge.

For example, search is currently quite limited. In the future we'd like to be able to search video automatically say for frames that contain a dog dancing. Less friviously, a medical researcher might wish to automatically find the comonalities in all patient's records for patients that have died of a particular condition.

Something that I think should be fully available and would be of tremendous benefit is the use of machine learning tools to help medical diagnosis. It's quite feasible to imagine that a good fraction of the medical knowledge of which symptoms are related to which diseases. Based on this we should never be in the situation of a doctor `missing' a diagnosis. Indeed, it's quite conceivable that the entire medical history of an individual should be signficantly able to help form a diagnosis.

Even in the developed world, transport systems are currently a real mess. I'd hope that machine learning would be able to provide some assistance there. Driverless cars are an obvious example, but certainly technology to avoid almost any serious car accident should be feasible. It's just incredible the number of people that are killed each year in car accidents -- surely machine learning should be able to help in almost eradicate such accidents.

This question originally appeared on Quora. - the knowledge sharing network where compelling questions are answered by people with unique insights. You can follow Quora on Twitter, Facebook, and Google+. More questions:​