Artificial Intelligence: Presentations
Search box is the most widely used input method for customers to obtain relevant products, but understanding customers’ queries is a complex task. In this paper we will describe the query understanding pipeline used for e-commerce. We go into the details on how phrase detection of queries is performed to understand the underlying concepts in the query instead of simply using bag of words.
Computer scientists have long drawn inspiration from neuroscience to produce computational tools. This presentation introduces neuroscience-inspired dynamic architectures (NIDAs), spiking neural networks embedded in a geometric space that have been successfully applied to applications in classification, control, and anomaly detection. A training method for NIDA networks based on evolutionary optimization is also introduced. A neuromorphic hardware implementation based on NIDA networks, Dynamic Adaptive Neural Network Array (DANNA), is discussed.
We propose a machine learning model for assessing the authoritativeness of news authors automatically. Authoritativeness of news authors refers to the ability of the author to provide substantiated evidence that reflects the quality of their writing. Our model is based on features extracted from the content of her history of articles and social feeds. We evaluate our model using a significant number of human assessment about the authoritativeness of authors.