Predictive Analytics Times
Predictive Analytics Times

4 weeks ago
Wise Practitioner – Predictive Analytics Interview Series: Arnab Chakrabarti at Hitachi America


By: Luba Gloukhova, Founding Chair, Deep Learning World

In anticipation of his upcoming conference presentation at Deep Learning World Las Vegas, June 16-20, 2019, we asked Arnab Chakrabarti, Senior Research Scientist at Hitachi America, a few questions about his work in deep learning. Catch a glimpse of his presentation, When is the Next Recession?: Forecasting with Recurrent Neural Networks​, and see what’s in store at the DLW conference in Las Vegas.

Q: In your work with deep learning, what do you model (i.e., what is the dependent variable, the behavior or outcome your models predict)?

A: The dependent variable is “real US GDP growth rate”.

Q: How does deep learning deliver value at your organization – what is one specific way in which model outputs actively drive decisions or operations?
A: Through predictive analytics. The application space can be finance, industrial AI, healthcare, energy or something else.

Q: Can you describe a quantitative result, such as the performance of your model or the ROI of the model deployment initiative?

A:  The proposed model leads to lower test error (MSE) than econometric models, human forecasters, as well as naively trained deep learning models.

Q: What surprising discovery or insight have you unearthed in your data?
A: We were able to train a deep learning model with just 40 years of GDP data sampled quarterly. The LSTM model’s power to generalize surprised us.

Q: What excites you most about the field of deep learning today?

A: Deep learning is enabling solutions that were not practical fifteen years ago.

Q: Sneak preview: Please tell us a take-away that you will provide during your talk at Deep Learning World.

A: We predict recessions with deep neural networks. Feature engineering plays an important role in improving the prediction, making it both more accurate and more robust – this is something we will discuss. The audience is often curious about the next recession, which is something we can discuss. Finally, there is an interesting insight on what the model actually does – does it predict recessions, or does it learn to reason like a human forecaster? Our results provide a clear answer to this last question.


Don’t miss Arnab’s presentation, When is the Next Recession?: Forecasting with Recurrent Neural Networks​, at DLW on Tuesday, June 18, 2019 from 5:10 to 5:30 PM. Click here to register for attendance.

By: Luba Gloukhova, Founding Chair, Deep Learning World

Leave a Reply