How Smart Are AI Chips, Really?

November 4, 2019 by Maya Posch 52 Comments

The best part about the term “Artificial Intelligence” is that nobody can really tell you what it exactly means. The main reason for this stems from the term “intelligence”, with definitions ranging from the ability to practice logical reasoning to the ability to perform cognitive tasks or dream up symphonies. When it comes to human intelligence, properties such as self-awareness, complex cognitive feats, and the ability to plan and motivate oneself are generally considered to be defining features. But frankly, what is and isn’t “intelligence” is open to debate.

What isn’t open to debate is that AI is a marketing goldmine. The vagueness has allowed for marketing departments around the world to go all AI-happy, declaring that their product is AI-enabled and insisting that their speech assistant responds ‘intelligently’ to one’s queries. One might begin to believe that we’re on the cusp of a fantastic future inhabited by androids and strong AIs attending to our every whim.

In this article we’ll be looking at the reality behind these claims and ponder humanity’s progress towards becoming a Type I civilization. But this is Hackaday, so we’re also going to dig into the guts of some AI chips, including the Kendryte K210 and see how the hardware of today fits into our Glorious Future. Continue reading “How Smart Are AI Chips, Really?” →

An Algorithm For De-Biasing AI Systems

October 21, 2019 by Sharon Lin 52 Comments

A fundamental truth about AI systems is that training the system with biased data creates biased results. This can be especially dangerous when the systems are being used to predict crime or select sentences for criminals, since they can hinge on unrelated traits such as race or gender to make determinations.

A group of researchers from the Massachusetts Institute of Technology (MIT) CSAIL is working on a solution to “de-bias” data by resampling it to be more balanced. The paper published by PhD students [Alexander Amini] and [Ava Soleimany] describes an algorithm that can learn a specific task – such as facial recognition – as well as the structure of the training data, which allows it to identify and minimize any hidden biases.

Testing showed that the algorithm minimized “categorical bias” by over 60% compared against other widely cited facial detection models, all while maintaining the same precision of detection. This figure was maintained when the team evaluated a facial-image dataset from the Algorithmic Justice League, a spin-off group from the MIT Media Lab.

The team says that their algorithm would be particularly relevant for large datasets that can’t easily be vetted by a human, and can potentially rectify algorithms used in security, law enforcement, and other domains beyond facial detection.

A Tamagotchi For WiFi Cracking

October 16, 2019 by Bryan Cockfield 8 Comments

OK, let’s start this one by saying that it’s useful to know how to break security measures in order to understand how to better defend yourself, and that you shouldn’t break into any network you don’t have access to. That being said, if you want to learn about security and the weaknesses within the WPA standard, there’s no better way to do it than with a tool that mimics the behavior of a Tamagotchi.

Called the pwnagotchi, this package of artificial intelligence looks for information in local WiFi packets that can be used to crack WPA encryption. It’s able to modify itself in order to maximize the amount of useful information it’s able to obtain from whatever environment you happen to place it in. As an interesting design choice, the pwnagotchi behaves like an old Tamagotchi pet would, acting happy when it gets the inputs it needs.

This project is beyond a novelty though and goes deep in the weeds of network security. If you’re at all interested in the ways in which your own networks might be at risk, this might be a tool you can use to learn a little more about the ways of encryption, general security, and AI to boot. Of course, if you’re new to the network security world, make sure the networks you’re using are secured at least a little bit first.

Thanks to [Itay] for the tip!

AI Makes Hyperbolic Brain Hats A Reality

September 14, 2019 by Jenny List 7 Comments

It isn’t often that the world of Hackaday intersects with the world of crafting, which is perhaps a shame because many of the skills and techniques of the two have significant overlap. Crochet for instance has rarely featured here, but that is about to change with [Janelle Shane]’s HAT3000 neural network trained to produce crochet hat patterns.

Taking the GPT-2 neural network trained on Internet text and further training it with a stack of crochet hat patterns, she was able to generate AI-designed hats which her friends on the Ravelry yarn forum set to crochet into real hats. It’s a follow-up to a previous knitting-based project, and instead of producing the hats you might expect it goes into flights of fancy. Some are visibly hat-like while others turn into avant-garde creations that defy any attempt to match them to real heads. A whole genre of hyperbolic progressions of crochet rows produce hats with organic folds that begin to resemble brains, and tax both the stamina of the person doing the crochet and their supply of yarn.

Perhaps most amusingly the neural network retains the ability to produce text, but when it does so it now inevitably steers the subject back to crochet hats. A Harry Potter sentence spawns a passage of something she aptly describes as “terrible crochet-themed erotica“, and such is the influence of the crochet patterns that this purple prose can even include enough crochet instructions to make them crochetable. It would be fascinating to see whether a similar model trained with G-code from Thingiverse would produce printable designs, what would an AI make with Benchy for example?

We’ve been entertained by [Janelle]’s AI work before, both naming tomato varieties, and creating pie recipes.

Thanks [Laura] for the tip.

Keep Pesky Cats At Bay With A Machine-Learning Turret Gun

July 12, 2019 by Dan Maloney 32 Comments

It doesn’t take long after getting a cat in your life to learn who’s really in charge. Cats do pretty much what they want to do, when they want to do it, and for exactly as long as it suits them. Any correlation with your wants and needs is strictly coincidental, and subject to change without notice, because cats.

[Alvaro Ferrán Cifuentes] almost learned this the hard way, when his cat developed a habit of exploring the countertops in his kitchen and nearly turned on the cooktop while he was away. To modulate this behavior, [Alvaro] built this AI Nerf turret gun. The business end of the system is just a gun mounted on a pan-tilt base made from 3D-printed parts and a pair of hobby servos. A webcam rides atop the gun and feeds into a PC running software that implements the YOLO3 localization algorithm. The program finds the cat, tracks its centroid, and swivels the gun to match it. If the cat stays in the no-go zone above the countertop for three seconds, he gets a dart in his general direction. [Alvaro] found that the noise of the gun tracking him was enough to send the cat scampering, proving that cats are capable of learning as long as it suits them.

We like this build and appreciate any attempt to bring order to the chaos a cat can bring to a household. It also puts us in mind of [Matthias Wandel]’s recent attempt to keep warm in his shop, although his detection algorithm was much simpler.

Continue reading “Keep Pesky Cats At Bay With A Machine-Learning Turret Gun” →

AI Recognizes And Locks Out Murder Cats

July 1, 2019 by Dan Maloney 56 Comments

Anyone with a cat knows that the little purring ball of fluff in your lap is one tiny step away from turning into a bloodthirsty serial killer. Give kitty half a chance and something small and defenseless is going to meet a slow, painful end. And your little killer is as likely as not to show off its handiwork by bringing home its victim – “Look what I did for you, human! Are you not proud?”

As useful as a murder-cat can be, dragging the bodies home for you to deal with can be – inconvenient. To thwart his adorable serial killer [Metric], Amazon engineer [Ben Hamm] turned to an AI system to lock his prey-laden cat out of the house. [Metric] comes and goes as he pleases through a cat flap, which thanks to a solenoid and an Arduino is now lockable. The decision to block entrance to [Metric] is based on an Amazon AWS DeepLens AI camera, which watches the approach to the cat flap. [Ben] trained three models: one to determine if [Metric] was in the scene, one to determine whether he’s coming or going, and one to see if he’s alone or accompanied by a lifeless friend, in which case he’s locked out for 15 minutes and an automatic donation is made to the Audubon Society – that last bit is pure genius. The video below is a brief but hilarious summary of the project for an audience in Seattle that really seems quite amused by the whole thing.

So your cat isn’t quite the murder fiend that [Metric] is? An RFID-based cat door might suit your needs better.

Continue reading “AI Recognizes And Locks Out Murder Cats” →

Artificial Neural Networks Help Hack Biological Counterparts

May 12, 2019 by Roger Cheng 16 Comments

Eyes are windows into the soul, the old saying goes. They are also pathways into the mind, as much of our brain is involved in processing visual input. This dedication to vision is partly why much of AI research is likewise focused on machine vision. But do artificial neural networks (ANN) actually work like the gray matter that inspired them? A recently published research paper (DOI: 10.1126/science.aav9436) builds a convincing argument for “yes”.

Neural nets were named because their organization was inspired by biological neurons in the brain. But as we learned more and more about how biological neurons worked, we also discovered artificial neurons aren’t very faithful digital copies of the original. This cast doubt whether machine vision neural nets actually function like their natural inspiration, or if they worked in an entirely different way.

This experiment took a trained machine vision network and analyzed its internals. Armed with this knowledge, images were created and tailored for the purpose of triggering high activity in specific neurons. These responses were far stronger than what occurs when processing normal visual input. These tailored images were then shown to three macaque monkeys fitted with electrodes monitoring their neuron activity, which picked up similarly strong neural responses atypical of normal vision.

Manipulating neural activity beyond their normal operating range via tailored imagery is the Hollywood portrayal of mind control, but we’re not at risk of input injection attacks on our brains. This data point gives machine learning researchers confidence their work still has relevance to biological source material, and neuroscientists are excited about the possibility of exploring brain functions without invasive surgical implants. Artificial neural networks could end up help us better understand what happens inside our brain, bringing the process full circle.

[via Science News]