Amazon Technologies, Inc.

Sling Media Pvt. Ltd.

Venkata Kishore Nandury

Ariya Rastrow

Andrew Oberlin

Pankaj Sitpure

Gurpreet Singh Chadha

Milind Mukesh Rao

Santosh Kumar Patro

Buddha Puneeth Nandanoor

Anand Mohan

Gopinath Chennupati

Prahalad Venkataramanan

A first neural network (NN) model may generate labels for training a second NN model. The second NN model may represent instances of a NN model operating on multiple different devices (e.g., decentralized user and/or edge devices). The system may include using a 'teacher' model to process data received by one or more of the devices to generate a labeled dataset. The system may use the labeled dataset and a “student” model to calculate gradient data for updating the student model. The student model may be the same or similar to NN model instances operating on the devices. The system may validate the updated student model to determine, for example, whether it exhibits improved performance when processing the newly received data and/or historical data. The system may distribute the validated update to the devices.

Continuous learning for machine learning models

Systems, methods, and computer-readable media are disclosed for identification of inaccurate addresses for package deliveries. In one embodiment, an example method may include determining a shipping address for a package that includes a product. A recurrent neural network may generate an address deliverability score by providing the shipping address as input. A set of corrective actions corresponding to the address deliverability score may be determined. The address deliverability score and the set of corrective actions may be presented. A selection of a corrective action from the set of corrective actions to implement prior to delivery of the package may be received.

Identification of inaccurate addresses for package deliveries

An adaptive gain control system and related operating method for digital audio samples is provided. The method is suitable for use with a digital media encoding system that transmits encoded media streams to a remotely-located presentation device such as a media player. The method begins by initializing the processing of a media stream. Then, the method adjusts the gain of a first set of digital audio samples in the media stream using a fast gain adaptation scheme, resulting in a first group of gain-adjusted digital audio samples having normalized volume during presentation. The method continues by adjusting the gain of a second set of digital audio samples in the media stream using a steady state gain adaptation scheme that is different than the fast gain adaptation scheme, resulting in a second group of gain-adjusted digital audio samples having normalized volume during presentation.

Adaptive gain control for digital audio samples in a media stream

Growing community of inventors

Bengaluru, India

Venkata Kishore Nandury