🎛 AI Accelerators And How Engineers Can Utilise Them
Understanding the role of AI accelerators in artificial intelligence gives an insight into its benefits in the world of electronics and computing
AI accelerators are hardware chips specifically designed to accelerate and facilitate machine learning computations. In the bid to utilising artificial intelligence to its full potential at minimal power requirements, AI accelerators can be incorporated in machines to improve basic machine computations, enhance system performance, etc.
To achieve this, AI accelerators have been utilised to accelerate the analysis of large linear algebra computations and parallelised them into smaller, often identical parts, some of which operate simultaneously.
Some applications of artificial intelligence include the Internet of Things, algorithms for guiding robotic decision-making, and other data-intensive tasks.
More information...
#articles #machine_learning #artificial_intelligence
Understanding the role of AI accelerators in artificial intelligence gives an insight into its benefits in the world of electronics and computing
AI accelerators are hardware chips specifically designed to accelerate and facilitate machine learning computations. In the bid to utilising artificial intelligence to its full potential at minimal power requirements, AI accelerators can be incorporated in machines to improve basic machine computations, enhance system performance, etc.
To achieve this, AI accelerators have been utilised to accelerate the analysis of large linear algebra computations and parallelised them into smaller, often identical parts, some of which operate simultaneously.
Some applications of artificial intelligence include the Internet of Things, algorithms for guiding robotic decision-making, and other data-intensive tasks.
More information...
#articles #machine_learning #artificial_intelligence
🎙 Google releases the source code for Lyra low bitrate speech codec
Google has now released the Lyra source code, written in C++ for optimal speed, efficiency, and interoperability and relying on both the Bazel build framework and the GoogleTest framework
Whereas traditional codecs are based on digital signal processing (DSP) techniques, the key advantage for Lyra comes from the ability of the generative model to reconstruct a high-quality voice signal.
More information...
#news #software #machine_learning
Google has now released the Lyra source code, written in C++ for optimal speed, efficiency, and interoperability and relying on both the Bazel build framework and the GoogleTest framework
Whereas traditional codecs are based on digital signal processing (DSP) techniques, the key advantage for Lyra comes from the ability of the generative model to reconstruct a high-quality voice signal.
More information...
#news #software #machine_learning
🧠 Flashlight: Fast and flexible machine learning in C++
Flashlight is a new open source machine learning (ML) library, written entirely in C++, that was built by FAIR to power groundbreaking research by enabling teams to rapidly and easily modify deep and ML frameworks to better fit their needs.
More information...
#articles #machine_learning #software #programming #artificial_intelligence #cpp #libraries
Flashlight is a new open source machine learning (ML) library, written entirely in C++, that was built by FAIR to power groundbreaking research by enabling teams to rapidly and easily modify deep and ML frameworks to better fit their needs.
More information...
#articles #machine_learning #software #programming #artificial_intelligence #cpp #libraries
This media is not supported in your browser
VIEW IN TELEGRAM
🤖 Apple harvesting robot plucks a piece of fruit every 7 seconds
Like many industries across the world, 🇦🇺 Australia's fruit-picking sector has been impacted heavily by the COVID-19 pandemic, creating a massive void in the labor market that would normally be filled by backpackers looking to make buck on the road.
Local researchers have cooked up a creative partial solution to this problem, developing a fruit-picking robot that can harvest apples from orchards at high speed.
The robot uses a mix of cameras and deep learning algorithms to scan the trees of an orchard and detect the pieces of fruit, which requires it to process information on their shape, orientation and the location of the stem-branch joint to minimize damage to the produce and the surrounding foliage.
More information...
#news #robots #machine_learning
Like many industries across the world, 🇦🇺 Australia's fruit-picking sector has been impacted heavily by the COVID-19 pandemic, creating a massive void in the labor market that would normally be filled by backpackers looking to make buck on the road.
Local researchers have cooked up a creative partial solution to this problem, developing a fruit-picking robot that can harvest apples from orchards at high speed.
The robot uses a mix of cameras and deep learning algorithms to scan the trees of an orchard and detect the pieces of fruit, which requires it to process information on their shape, orientation and the location of the stem-branch joint to minimize damage to the produce and the surrounding foliage.
More information...
#news #robots #machine_learning
📃 ARM vs RISC-V Vector Extensions
A comparison of the RISC-V vector extension (RVV) and ARM scalable vector extension (SVE/SVE2)
Microprocessor with vector instructions is going to be the big thing for the future. Why? Because self-driving, speech recognition, image recognition are all based on machine learning and machine learning is all about matrices and vectors.
But that is not the only reason. We have been banging our heads in the wall trying to eek out more performance for years ever since we semi-officially declared Moore’s laws to be over. In the golden old days of microprocessor design, we could simply double the clock frequency of the CPU each year and boom everybody was happy. That wonderful old trick is over.
Today we play a thousand different clever games to eek out more performance whether that is through adding more CPU cores, adding out-of-order execution, more advance branch predictors or SIMD instructions.
More information...
#articles #electronics #machine_learning
A comparison of the RISC-V vector extension (RVV) and ARM scalable vector extension (SVE/SVE2)
Microprocessor with vector instructions is going to be the big thing for the future. Why? Because self-driving, speech recognition, image recognition are all based on machine learning and machine learning is all about matrices and vectors.
But that is not the only reason. We have been banging our heads in the wall trying to eek out more performance for years ever since we semi-officially declared Moore’s laws to be over. In the golden old days of microprocessor design, we could simply double the clock frequency of the CPU each year and boom everybody was happy. That wonderful old trick is over.
Today we play a thousand different clever games to eek out more performance whether that is through adding more CPU cores, adding out-of-order execution, more advance branch predictors or SIMD instructions.
More information...
#articles #electronics #machine_learning
🧠 Free Machine Learning course by Andrew Ng
This course provides a broad introduction to machine learning, datamining, and statistical pattern recognition. Topics include:
▫️ Supervised learning (parametric/non-parametric algorithms, support vector machines, kernels, neural networks).
▫️ Unsupervised learning (clustering, dimensionality reduction, recommender systems, deep learning).
▫️ Best practices in machine learning (bias/variance theory; innovation process in machine learning and AI).
More information:
Youtube playlist with lectures
Course on Coursera
Matlab implementation
Python implementation:
Part 1
Part 2
Part 3
Part 4
#courses #machine_learning #python #artificial_intelligence #programming
This course provides a broad introduction to machine learning, datamining, and statistical pattern recognition. Topics include:
▫️ Supervised learning (parametric/non-parametric algorithms, support vector machines, kernels, neural networks).
▫️ Unsupervised learning (clustering, dimensionality reduction, recommender systems, deep learning).
▫️ Best practices in machine learning (bias/variance theory; innovation process in machine learning and AI).
More information:
Youtube playlist with lectures
Course on Coursera
Matlab implementation
Python implementation:
Part 1
Part 2
Part 3
Part 4
#courses #machine_learning #python #artificial_intelligence #programming
📰 Embedded Vision Sees a Bright Future
Driven by advances in sensors, processors and software, embedded vision is going everywhere – from agriculture to factories, and from autonomous vehicles to professional sports. Even the Covid-19 pandemic has served to accelerate its deployment with vision systems being used in applications like public surveillance, health and safety inspection.
Artificial intelligence (AI) is gaining momentum in embedded vision and image processing applications as developers increasingly apply deep learning and neural networks to improve object detection and classification.
With an increased ease-of-use, lower prices, and smaller devices that fit into existing machineries, embedded vision will be more accessible for smaller companies that haven’t used embedded vision before.
More information...
#news #machine_learning #artificial_intelligence
Driven by advances in sensors, processors and software, embedded vision is going everywhere – from agriculture to factories, and from autonomous vehicles to professional sports. Even the Covid-19 pandemic has served to accelerate its deployment with vision systems being used in applications like public surveillance, health and safety inspection.
Artificial intelligence (AI) is gaining momentum in embedded vision and image processing applications as developers increasingly apply deep learning and neural networks to improve object detection and classification.
With an increased ease-of-use, lower prices, and smaller devices that fit into existing machineries, embedded vision will be more accessible for smaller companies that haven’t used embedded vision before.
More information...
#news #machine_learning #artificial_intelligence
Discussing the trajectory of AI-powered military technology
OSMpod
🎙 On the Radar Podcast: Discussing the trajectory of AI-powered military technology
Artificial intelligence (AI) is one of the hottest topics in modern technology, and defense electronics are no exception.
Base-patrolling robot dogs and algorithms designed to understand complex combat scenarios are currently in development, but that's just the beginning. In the debut episode of On the Radar, Emma Helfrich and John McHale of Military Embedded Systems discuss the current state of military AI and machine learning (ML), how these advancements are being financed, and the obstacles that stand in innovation's way.
#podcasts #machine_learning #artificial_intelligence
Artificial intelligence (AI) is one of the hottest topics in modern technology, and defense electronics are no exception.
Base-patrolling robot dogs and algorithms designed to understand complex combat scenarios are currently in development, but that's just the beginning. In the debut episode of On the Radar, Emma Helfrich and John McHale of Military Embedded Systems discuss the current state of military AI and machine learning (ML), how these advancements are being financed, and the obstacles that stand in innovation's way.
#podcasts #machine_learning #artificial_intelligence
🍅 FarmBot - open-source farming robot
FarmBot is an open source precision agriculture CNC farming project consisting of a Cartesian coordinate robot farming machine, software and documentation including a farming data repository. This interesting start-up focuses on bringing private automated farms to its clients. Robot uses machine learning algorithms for plants' analysis and takes actions depending on its results.
The FarmBot Genesis is able to plant over 30 different crops including potatoes, peas, squash, artichokes and chard in an area of 2.9 meters × 1.4 meters with a maximum plant height of 0.5 meters.
The Farmbot Genesis can perform almost all processes prior to harvesting including sowing, mechanical weed control and watering. It requires electricity, an internet connection and water supply which can be provided using off grid solutions including a water barrel to collect rain and a solar panel and battery to provide electricity.
More information...
#robots #machine_learning
FarmBot is an open source precision agriculture CNC farming project consisting of a Cartesian coordinate robot farming machine, software and documentation including a farming data repository. This interesting start-up focuses on bringing private automated farms to its clients. Robot uses machine learning algorithms for plants' analysis and takes actions depending on its results.
The FarmBot Genesis is able to plant over 30 different crops including potatoes, peas, squash, artichokes and chard in an area of 2.9 meters × 1.4 meters with a maximum plant height of 0.5 meters.
The Farmbot Genesis can perform almost all processes prior to harvesting including sowing, mechanical weed control and watering. It requires electricity, an internet connection and water supply which can be provided using off grid solutions including a water barrel to collect rain and a solar panel and battery to provide electricity.
More information...
#robots #machine_learning
👍2
🧠 Dlib - modern open-source machine learning toolkit written in C++
Dlib is a toolkit containing machine learning algorithms and tools for creating complex software to solve real world problems.
It is used in both industry and academia in a wide range of domains including robotics, embedded devices, mobile phones, and large high performance computing environments. Dlib's open source licensing allows you to use it in any application, free of charge.
Main features:
▫️ Documentation - provides complete and precise documentation for every class and function
▫️ High Quality Portable Code
▫️ Huge Variety of Machine Learning Algorithms
▫️ Image Processing capabilities, such as high quality face recognition
▫️ Graphical User Interfaces
Can be easily implemented on microcontrollers, such as Raspberry Pi - example: video, github.
More information...
#software #machine_learning
Dlib is a toolkit containing machine learning algorithms and tools for creating complex software to solve real world problems.
It is used in both industry and academia in a wide range of domains including robotics, embedded devices, mobile phones, and large high performance computing environments. Dlib's open source licensing allows you to use it in any application, free of charge.
Main features:
▫️ Documentation - provides complete and precise documentation for every class and function
▫️ High Quality Portable Code
▫️ Huge Variety of Machine Learning Algorithms
▫️ Image Processing capabilities, such as high quality face recognition
▫️ Graphical User Interfaces
Can be easily implemented on microcontrollers, such as Raspberry Pi - example: video, github.
More information...
#software #machine_learning
📟 "Listening Temperature" with TinyML
Teaching Arduino to hear the temperature difference using Edge Impulse Studio
Firstly, author acquire the sound of pouring water and then uploades it to Edge Impulse Studio, where the data gets preprocessed. Then he trains the Neural Network (NN), tests it and deploys to an MCU (Arduino Nano) for real physical test.
To do so, author uses Edge Optimized Neural (EON) Compiler, which permits to run of neural networks in 25-55% less RAM, and up to 35% less flash while retaining the same accuracy compared to TensorFlow Lite for Microcontrollers.
More information...
#projects #machine_learning #artificial_intelligence #arduino
Teaching Arduino to hear the temperature difference using Edge Impulse Studio
Firstly, author acquire the sound of pouring water and then uploades it to Edge Impulse Studio, where the data gets preprocessed. Then he trains the Neural Network (NN), tests it and deploys to an MCU (Arduino Nano) for real physical test.
To do so, author uses Edge Optimized Neural (EON) Compiler, which permits to run of neural networks in 25-55% less RAM, and up to 35% less flash while retaining the same accuracy compared to TensorFlow Lite for Microcontrollers.
More information...
#projects #machine_learning #artificial_intelligence #arduino
📟 Open-source plastic scanner
Towards a simple device that can identify the five most common plastics
Plastic pollution is a well-known problem worldwide, and is still growing. It negatively affects humans and wildlife through animal death, groundwater pollution and incorporation of micro plastics in our digestive system.
Discrete near-infrared spectroscopy makes it possible to identify over 75% of all plastic used in everyday life.
This ecosystem consists of:
▫️ A breakout board that combines all components to emit and sense infrared light on a small printed circuit board with standard communication protocols.
▫️ A handheld scanner that integrates the breakout board into real-world applications. It enables local machine-learning processing of the sample, all in a compact form factor.
▫️ Software that communicates with the hardware and implements machine learning for optimal and quick prediction of plastic types.
More information...
#projects #machine_learning
Towards a simple device that can identify the five most common plastics
Plastic pollution is a well-known problem worldwide, and is still growing. It negatively affects humans and wildlife through animal death, groundwater pollution and incorporation of micro plastics in our digestive system.
Discrete near-infrared spectroscopy makes it possible to identify over 75% of all plastic used in everyday life.
This ecosystem consists of:
▫️ A breakout board that combines all components to emit and sense infrared light on a small printed circuit board with standard communication protocols.
▫️ A handheld scanner that integrates the breakout board into real-world applications. It enables local machine-learning processing of the sample, all in a compact form factor.
▫️ Software that communicates with the hardware and implements machine learning for optimal and quick prediction of plastic types.
More information...
#projects #machine_learning
👍3
This media is not supported in your browser
VIEW IN TELEGRAM
🧠 Self-Parking Car in 500 Lines of Code
In this article, we'll train the car to do self-parking using a genetic algorithm. On the ≈40th generation the cars start learning what the self-parking is and start getting closer to the parking spot.
You may launch the Self-parking Car Evolution Simulator to see the evolution process directly in your browser. The simulator gives you the following opportunities:
▫️ You may train the cars from scratch and adjust genetic parameters by yourself.
▫️ You may see the trained self-parking cars in action.
▫️ You may also try to park the car manually.
The genetic algorithm for this project is implemented in TypeScript. The full genetic source code will be shown in this article, but you may also find the final code examples in the Evolution Simulator repository.
More information...
#articles #programming #machine_learning
In this article, we'll train the car to do self-parking using a genetic algorithm. On the ≈40th generation the cars start learning what the self-parking is and start getting closer to the parking spot.
You may launch the Self-parking Car Evolution Simulator to see the evolution process directly in your browser. The simulator gives you the following opportunities:
▫️ You may train the cars from scratch and adjust genetic parameters by yourself.
▫️ You may see the trained self-parking cars in action.
▫️ You may also try to park the car manually.
The genetic algorithm for this project is implemented in TypeScript. The full genetic source code will be shown in this article, but you may also find the final code examples in the Evolution Simulator repository.
More information...
#articles #programming #machine_learning
👍1
🧠 Popular ML algorithms (MATLAB, Octave, Python)
A perfect compilation of code examples of the most popular ML algorithms for understanding its work.
The purpose of these repositories was not to implement machine learning algorithms using 3rd party libraries "one-liners" but rather to practice and to better understand the mathematics behind each algorithm.
More information:
Python Repository
MATLAB/Octave Repository
#machine_learning #artificial_intelligence #matlab #python
A perfect compilation of code examples of the most popular ML algorithms for understanding its work.
The purpose of these repositories was not to implement machine learning algorithms using 3rd party libraries "one-liners" but rather to practice and to better understand the mathematics behind each algorithm.
More information:
Python Repository
MATLAB/Octave Repository
#machine_learning #artificial_intelligence #matlab #python
👍3
🤖 Stereo Vision: Making a Depth Map from scratch
Imagine you need to create an environmental garbage-collection robot. However, you do not want to invest heavily in LiDAR technology to determine how far the trash is. This is where Stereo Vision comes into play.
Using two regular pin-hole cameras to not only detect depth but also use the same camera for other Machine Learning tasks can be a huge cost-saving factor.
This project is a great way for beginners to step into the field of not just Computer Vision but Computer Science as a whole, picking up skills in Python and OpenCV.
To begin on this project, you'll need a few essentials:
▫️ Jetson Nano Developer Kit B01
▫️ Two Raspberry Pi cameras
▫️ Interest!
More information:
Part 1: Introduction
Part 2: Depth Map
Part 3: Adding Object Detection
#articles #robots #machine_learning #python
Imagine you need to create an environmental garbage-collection robot. However, you do not want to invest heavily in LiDAR technology to determine how far the trash is. This is where Stereo Vision comes into play.
Using two regular pin-hole cameras to not only detect depth but also use the same camera for other Machine Learning tasks can be a huge cost-saving factor.
This project is a great way for beginners to step into the field of not just Computer Vision but Computer Science as a whole, picking up skills in Python and OpenCV.
To begin on this project, you'll need a few essentials:
▫️ Jetson Nano Developer Kit B01
▫️ Two Raspberry Pi cameras
▫️ Interest!
More information:
Part 1: Introduction
Part 2: Depth Map
Part 3: Adding Object Detection
#articles #robots #machine_learning #python
👍3
📃 How to quickly deploy TinyML on MCUs?
This article will provide you with an introduction to ML on microcontrollers. This topic is also known as tiny machine learning (TinyML). Get ready for losing against an ESP-EYE at rock, paper, scissors.
You will learn about:
▫️ data collection and processing
▫️ how to design and train an AI
▫️ how to get it running on the MCU
This example provides you with all you need to do your own TinyML project, from start to end.
More information...
#articles #machine_learning #programming #python
This article will provide you with an introduction to ML on microcontrollers. This topic is also known as tiny machine learning (TinyML). Get ready for losing against an ESP-EYE at rock, paper, scissors.
You will learn about:
▫️ data collection and processing
▫️ how to design and train an AI
▫️ how to get it running on the MCU
This example provides you with all you need to do your own TinyML project, from start to end.
More information...
#articles #machine_learning #programming #python
👍8
This media is not supported in your browser
VIEW IN TELEGRAM
📟 TinyML-CAM pipeline enables 80 FPS image recognition on ESP32 using just 1 KB RAM
The challenge with TinyML is to extract the maximum performance/efficiency at the lowest footprint for AI workloads on microcontroller-class hardware. The TinyML-CAM pipeline, developed by a team of machine learning researchers in Europe, demonstrates what’s possible to achieve on relatively low-end hardware with a camera.
Most specifically, they managed to reach over 80 FPS image recognition on the sub-$10 ESP32-CAM board with the open-source TinyML-CAM pipeline taking just about 1KB of RAM. It should work on other MCU boards with a camera, and training does not seem complex since we are told it takes around 30 minutes to implement a customized task.
More information:
The main article
The GitHub repository
#projects #machine_learning #esp32
The challenge with TinyML is to extract the maximum performance/efficiency at the lowest footprint for AI workloads on microcontroller-class hardware. The TinyML-CAM pipeline, developed by a team of machine learning researchers in Europe, demonstrates what’s possible to achieve on relatively low-end hardware with a camera.
Most specifically, they managed to reach over 80 FPS image recognition on the sub-$10 ESP32-CAM board with the open-source TinyML-CAM pipeline taking just about 1KB of RAM. It should work on other MCU boards with a camera, and training does not seem complex since we are told it takes around 30 minutes to implement a customized task.
More information:
The main article
The GitHub repository
#projects #machine_learning #esp32
👍15
🧠 Fido – light-weight C++ machine learning library for embedded electronics
Fido is an light-weight, highly modular C++ machine learning library for embedded electronics and robotics. Fido is especially suited for robotic and embedded contexts, as it is written in C++ with minimal use of the standard library, comes packaged with a robotic simulator, and provides and easy interface in which to write robotic drivers.
More information...
#libraries #programming #machine_learning
Fido is an light-weight, highly modular C++ machine learning library for embedded electronics and robotics. Fido is especially suited for robotic and embedded contexts, as it is written in C++ with minimal use of the standard library, comes packaged with a robotic simulator, and provides and easy interface in which to write robotic drivers.
More information...
#libraries #programming #machine_learning
👍12👌1
Embedded Systems
📰 A company's new robot can change from four wheel drive to bipedal in seconds A company called Swiss-Mile has created an innovative robot that combines legs and wheels to create, what they hope, will be the most versatile last-mile delivery robot on the…
Zürich-based Swiss-Mile has enhanced the ANYmal quadruped robot with powered wheels for speed and efficiency, enabling it to handle curbs and stairs. The robot has been taught to stand up, facilitating more versatile movement and manipulation using its wheel-hand-leg-arms. Researchers at ETH Zurich's Robotic Systems Lab employed curiosity-driven learning, a form of reinforcement learning, to train the robot in tasks like opening doors and placing packages into boxes.
The robot's curiosity about specific elements, such as the door handle or package motion, helps it explore and innovate solutions independently. The use of sparse rewards in training results in diverse and reliable behaviors, allowing the robot to recover from real-world mistakes. While relying on visual code-based systems like AprilTags, Swiss-Mile aims to further develop the robot's capabilities, indicating potential practical applications for the ANYmal.
#news #robots #machine_learning
Please open Telegram to view this post
VIEW IN TELEGRAM
Please open Telegram to view this post
VIEW IN TELEGRAM
👍13👌1
BitNetMCU presents an exciting journey into the fusion of machine learning and embedded systems, offering valuable insights into the implementation of neural networks even on the most modest piece of hardware.
Features of the project:
To sum up, this project provides a gateway to deeper understanding in electronics, AI, and machine learning, empowering individuals to create innovative solutions for real-world challenges in IoT, edge computing, and beyond.
More information:
#projects #libraries #artificial_intelligence #machine_learning
Please open Telegram to view this post
VIEW IN TELEGRAM
👍9👌2