Artificial Intelligence models and programmes mimic the functioning of human brains, helping machines learn make decisions in a more-like manner. But how exactly they arrive at decisions is unknown.
In order to understand this, a group of researchers at the Indian Institute of Technology (IIT-Hyderabad) have developed a method by which the inner workings of Artificial Intelligence models can be understood in terms of causal attributes.
This finding assumes significance in the wake of emergence of regulations such as General Data Protection Regulation (GDPR) that requires organisations to explain the decisions made by machine learning methods.
Modern Artificial Neural Networks, also called Deep Learning (DL), have increased tremendously in complexity such that machines can train themselves to process and learn from data that has been fed to them. They almost match human performance in many tasks.
“Transparency and understandability of the workings of DL models are gaining importance as discussions around the ethics of Artificial intelligence grows,” Vineeth N. Balasubramanian, Associate Professor (Department of Computer Science and Engineering) at IIT- Hyderabad, said.
He along with his students Aditya Chattopadhyay, Piyushi Manupriya, and Anirban Sarkar have worked on the project. Their work has recently been published in the Proceedings of 36th International Conference on Machine Learning.
“There are challenges to be met as the achievements in the field of AI and machine learning have wowed everyone,” he said.
“A key bottleneck in accepting such Deep Learning models in real-life applications, especially risk-sensitive ones, is the ‘interpretability problem,” he said.
“The DL models, because of their complexity and multiple layers, become virtual black boxes that cannot be deciphered easily.
This makes troubleshooting difficult, if not impossible,” Balasubramanian explained.
The IIT Hyderabad team approached this problem with ANN architectures using causal inference with what is known in the field as a ‘Structural Causal Model.’
“We have proposed a new method to compute the Average Causal Effect of an input neuron on an output neuron. It is important to understand which input parameter is ‘causally’ responsible for a given output,” he pointed out.
For example in the field of medicine, how does one know which patient attribute was causally responsible for the heart attack? “Our (IIT Hyderabad researchers’) method provides a tool to analyse such causal effects,” he added