Amazon

The Amazon Robotics Journey To ‘Proteus’, The Company’s First Fully Automated Mobile Robot

The past ten years have sped by, but Amazon’s desire to improve their workers' and customers' experiences has remained constant. A committed group of...

Amazon AI Researchers Open-Source ‘Syne Tune’: A Novel Python Library For Distributed HPO With An Emphasis On Enabling Reproducible Machine Learning Research

Deep learning models with billions of parameters are trained through gradient-based stochastic optimization, thanks to powerful algorithms, systems, and hardware advancements. These algorithms include...

New Amazon AI Study Introduces Pyramid-BERT To Reduce Complexity via Successive Core-set based Token Selection

This Article is written as a summay by Marktechpost Staff based on the paper 'Pyramid-BERT: Reducing complexity via successive core-set based token selection'. All...

Amazon AI Researchers Proposed ‘DQ-BART’: A Jointly Distilled And Quantized BART Model That Achieves 16.5x Model Footprint Compression Ratio

This Article is written as a summay by Marktechpost Staff based on the paper 'DQ-BART: Efficient Sequence-to-Sequence Model via Joint Distillation and Quantization'. All...

Amazon AI Researchers Propose A New Model, Called RescoreBERT, That Trains A BERT Rescoring Model With Discriminative Objective Functions And Improves ASR Rescoring

This Article is written as a summay by Marktechpost Staff based on the Research Paper 'RESCOREBERT: DISCRIMINATIVE SPEECH RECOGNITION RESCORING WITH BERT'. All Credit...

Amazon Researchers Develop A New Way To Rewrite Database Queries To Eliminate Redundancies

As databases grow in size, queries become slower. Database queries frequently contain multiple repetitive procedures that can be eliminated. To discover a full name...

AI Researchers At Amazon Develop GIT: Generative Insertion Transformer For Controlled Data Generation Via Insertion Operations

This Article Is Based On The Research 'Controlled Data Generation via Insertion Operations for NLU'. All Credit For This Research Goes To The Researchers...

Researchers From Cambridge And Amazon Propose ‘TRANS-ENCODER’: An Unsupervised Approach of Training Bi- And Cross-Encoders For Sentence-Pair Tasks

This Article Is Based On The Research Article 'TRANS-ENCODER: UNSUPERVISED SENTENCE-PAIR MODELLING THROUGH SELF- AND MUTUAL-DISTILLATIONS'. All Credit For This Research Goes To The...

Amazon Researchers Propose ‘Cold Brew’: A Teacher-Student Distillation Approach To Address The SCS And Noisy-Neighbor Challenges For Graph Neural Networks (GNNs)

This Article Is Based On The Research Paper 'Cold Brew: Distilling graph node representations with incomplete or missing neighborhoods'.. All Credit For This Research...

Amazon Kickstarts Natural Language Understanding By Open-Sourcing ‘MASSIVE’ Speech Dataset

This Article Is Based On The Amazon article 'Amazon releases 51-language dataset for language understanding'. All Credit For This Research Goes To The Amazon...

Latest Paper From Amazon AI Research Analyzes And Explains The Challenges And Developments in The Field Of Federated Learning

This article summary is based on the research paper from Amazon: 'Federated learning challenges and opportunities: An outlook' All credits for this research goes to...

Latest AI Research at Amazon Improves Forecasting by Learning the Quantile Functions

This research summary is based on the papers 'Learning quantile functions without quantile crossing for distribution-free time series forecasting' and 'Multivariate quantile function forecaster' Please...

Recent articles