Pinar Yilmaz

Organizations must be able to explain to their customers and internal teams why they made a certain decision based on an ML model’s prediction. Explaining ML models and understanding the reasoning behind the predictions can be difficult but is critical for responsible ML use.

State-of-the-art deep learning (DL) models for natural language processing and computer vision tasks are complex, multi-layered neural networks with billions of parameters that can take thousands of GPU hours to train. Long model training time slows down experimentation and hinders innovation.

Explaining ML models and understanding the reasoning behind the predictions they make is often difficult, but it is important to support responsible use of ML. Trained models may weigh some features more heavily than others when generating predictions.

Machine learning (ML) models may generate predictions that are not fair, whether because of biased data, a model that contains bias, or bias that emerges over time as real-world conditions change. Likewise, closed-box ML models are opaque, making it difficult to explain to internal stakeholders, auditors, external regulators, and customers alike why models make predictions both overall and for...