Artificial Intelligence (AI) is changing every aspect of consumer product design. Manufacturers are discovering new ways to deploy AI into their products to differentiate in the market with new capabilities, improved efficiency, and reduced operating costs. However, the extreme processing requirements of AI have led manufacturers to implement AI either on-device using a high-cost, AI-capable chip, with a connection to cloud-based AI infrastructure, or both.
These approaches incur design and operating costs that have priced AI out of all but the highest-end products in markets such as white goods/home appliances and other consumer electronics applications. To make AI accessible to cost-sensitive mass markets, we need to bring costs down. This is where sparse AI technology can help. By optimizing AI inference processing by up to 100 times, sparse AI enables developers to implement complex, deep-learning-based AI models using low-cost AI MCU silicon without adversely impacting speed, efficiency, memory footprint, or performance. This article will explore sparse AI and how manufacturers can optimize AI inferencing to reduce cloudbased dependency and infrastructure or even implement powerful AI-based capabilities completely on-device.
SPARSE AI
AI models can be complex and require extensive processing resources and memory. In high-end AI systems, a specialized (and expensive) processor like a GPU runs AI model inferencing on-device. Alternatively, many systems take a cloud-based approach where data is collected on-device and sent to a server for processing.
Denne historien er fra September 2024-utgaven av Circuit Cellar.
Start din 7-dagers gratis prøveperiode på Magzter GOLD for å få tilgang til tusenvis av utvalgte premiumhistorier og 9000+ magasiner og aviser.
Allerede abonnent ? Logg på
Denne historien er fra September 2024-utgaven av Circuit Cellar.
Start din 7-dagers gratis prøveperiode på Magzter GOLD for å få tilgang til tusenvis av utvalgte premiumhistorier og 9000+ magasiner og aviser.
Allerede abonnent? Logg på
Bringing Cost-Effective, On-Device AI to Home Appliances - Sparse Weights and Interactions Negate GPUs and Cloud Computing
GPU silicon and cloud computing infrastructure are too costly for mass markets devices like refrigerators and washers. By deploying compute resources only to the necessary parts of AI inference, Sparse AI allows product designers to practically incorporate new AI features like natural voice interfaces into their consumer offerings without breaking the bank or electric bill surprises.
Thin Film Transistor LED Displays - Visual I/O
To add a resistive touchscreen on top of a TFT display, Jeff explores the technology offerings and libraries needed to accomplish his task. Calibrating the touchscreen's computed coordinates with the actual pixel coordinates of the display proved difficult with a modular framework but separating functions allows easy expansion.
Mobile APP Development with React Native
This month, Bob continues his series on mobile app development from an embedded designer's perspective. He programs a React Native app and he offers guidelines for picking a tutorial for React Native. He further explores how it handles concurrency and asynchronous programming. Bob expected a few nuances during the process but to say there were more than a few would be an understatement.
Power Analysis Over JTAG Ports: Hidden Debug Dangers - Block Side-Channel Analysis Attacks
Small changes in the phase of clock signals can encode power leakages. An attacker can use standard interfaces such as the JTAG port to drive clocks across targets, using them as measurement techniques for sidechannel analysis attacks despite limited physical access. This article demonstrates how you can test devices for vulnerability and how to modify devices to prevent attacks.
The Blues Wireless Notecard-LoRa
After having built a number of Wi-Fi IoT gadgets for the home, Brian became curious to see how LoRaWAN-based IoT devices would work. He decided to investigate the Notecard LoRa IoT modules made by Blues Wireless.
IoT Protection with MCU Security - Ward Off Attacks with Trust Zones, Encryption, Secure Keys and Tamper Detection
Security breaches jumped 107 percent as malevolent forces increasingly focused on IoT devices for side attacks, IP and data theft. Semiconductor vendors supplying on-chip cryptography, memory protection, unclonable technology, random number generation and other measures hope to shut the door on attackers.
Fill AI with Computer Vision - On the Leading Edge of Adoption in Industrial IoT
In just about any factory, production lines hum along like a symphonic orchestra. However, makers of artificial intelligence hardware are pushing transformative technology with mind-boggling productivity gains to unlock unrealized potential. Just how fast will AI seamlessly integrate into the means of production?
No Blues with Bluetooth! GATT Explained - Part 5 - Understanding Higher Protocol Layers Provides Interoperability Help
After exploring low-level layers of Bluetooth Low Energy in previous editions, and establishing connections in Part 4, Robert examines Attribute Protocol and Generic Attribute Profile or GATT in Part 5. These higher layer protocols provide developers with greater opportunity for interoperability with third-party applications or products.
Retro-style Miniature Sports Scoreboard - RGB LED Matrix and Raspberry Pi Provide In-Game Updates
This Cornell University graduate describes how he built a miniature, retrostyle scoreboard that pulls and displays live college and professional sports scores from the ESPN API-bringing the excitement of being at a sporting event to your living room. Pre-game information, in-game scores, and post-game results are displayed on a built-in PiTFT touchscreen.
Exploring Infrasound - Tornado Early Warning Detection Project Using Raspberry Pi Pico and a Python Supervisory Program
Inspired by the possibility of building an early warning system for tornadoes, Dev developed an infrasound detector. Tornadoes are rare in his area, but his detector found other infrasound activity using sensor circuitry based on a Raspberry Pi Pico, combined with a Python supervisory program that stores data on a computer for analysis.
