Google’s Edge TPU Machine Learning Chip Helps In Automating Quality Control For IoT Applications

 

The pioneer in the internet-related services and products: Google has unveiled a new tensor processing unit (TPU) called Edge TPU at the Google Cloud Next conference in San Francisco. The novel chip is designed to run TensorFlow Lite machine learning models on the Internet of Things-based (IoT-based) devices.

 

 

Two years ago, Google launched its TPU, a specialized chip that performs simplified tasks of artificial intelligence (AI). However, the newly-launched Edge TPU is carefully designed to carry out the so-called “inference tasks”, which is the part of machine learning where an algorithm performs the trained tasks such as identifying an object in the picture. Google’s server-based TPUs were optimized for the training part of this process while Edge TPUs will perform inference.

 

Google announced that these new chips are designed to be used only in enterprise tasks such as automating quality control checks at factories. Automation in quality control has a number of advantages over usage of hardware that sends the collected data over the internet for analysis. Machine learning chips are more secure and require less downtime while delivering results quickly.

 

Google’s machine learning chips can ensure that different parts of IoT applications talk to one another seamlessly, making it easier for customers to stay in the company’s ecosystem.

 

 

Injong Rhee, Google Cloud’s vice president of IoT, described the Edge TPU as a “purpose-built ASIC chip designed to run TensorFlow Lite ML models at the edge” in a blog post. Rhee stated

Edge TPUs are designed to complement our Cloud TPU offering, so you can accelerate ML training in the cloud, then have lightning-fast ML inference at the edge. Your sensors become more than data collectors — they make local, real-time, intelligent decisions.”

 

Google has announced to make Edge TPUs available as a development kit so that the customers can test out the hardware capability. This kit will include a system on module (SOM) containing the Edge TPU, a Microchip secure element, an NXP CPU, and Wi-Fi functionality. Moreover, it can be connected to a computer via USB or a PCI Express expansion slot. However, customers will have to apply for an access.

 

Google is not the only company that designs machine learning chips. ARM, Qualcomm, and Mediatek also design their dedicated AI chips. Moreover, GPU manufactured by NVidia is still the leader in the market. This increased competition has boosted the growth of the machine learning chips market.

 

 

According to Allied Market Research, the global machine learning chip market was valued at $2.43 billion in 2017 and is projected to reach $37.85 billion by 2025, registering a CAGR of 40.8% through 2025. The report includes an in-depth analysis of several market strategies adopted by market hulks such as Google, Intel, Nvidia, and Qualcomm.

 

Get your copy of What is Bitcoin

This e-book on Amazon explains what Bitcoin is, it explains that Bitcoin (BTC) is a virtual currency, digital, not physical, and independent of banks. Useful links and resources for the newbie and advanced Bitcoiner or cryptocurrency enthusiast.

 

 

Please leave your questions and comments below:

 

Download the free Bit-Media App from the Google Play or Apple stores.

Click below and get your FREE BIT-MEDIA APP

Are you a Bitcoin Believer? Test your belief here - The Bitcoin Believers Business Manifesto

BEZY, a Bold Vision for Empowerment and Wealth Creation

EOT, quietly creating the Economy of Things