Revolutionising Automated Driving Systems with Safety Pool
Today, Deepen AI and WMG, in collaboration with the University of Warwick, the launch of the Scenario Database. The database is long-awaited and will provide a variety of driving scenarios that can be leveraged by academia, governments, and industries to both test and certify upcoming Automated Driving Systems (ADSs) and Advanced Driver Assistance Systems (ADAS). The system will also help in the creation of policies and regulatory guidelines as the world continues to move towards automated driving norms and AI-controlled vehicles.
A Brief Factfile
- It’s generally accepted that autonomous vehicles have to be tested for a minimum of 11 billion miles before they can be deemed ‘road-ready’.
- To make 11 billion miles a viable reality, it is necessary that some of these miles are performed in virtual road scenarios ─ which Deepen AI and WMG at the University of Warwick have created a global database for.
- The Safety Pool™ Scenario Database will not only provide insights into the safety readiness of ADAS and ADS but also assist in hastening the global adoption of autonomous vehicles, which is classed as a necessity in our bid to outpace the ruin of Earth.
Michelle Avary of the shared her thoughts on the announcement this morning, stating that "We are thrilled to work closely with Deepen AI and WMG, University of Warwick to launch the Safety Pool™ Scenario Database. We believe Safety Pool™ is going to play a crucial role in standardising and bringing transparency to ADSs certification globally. We look forward to partnering with countries to adopt ADS certification frameworks based on Safety Pool™ Scenario Database.”
“The Safety Pool™ Scenario Database lays a key foundation stone for autonomous vehicle safety quantification," said Mohammad Musa, CEO & Co-founder of . "We are working closely with governments across the world to create a framework for ADS certification that will take OEMs one giant step closer to getting autonomous vehicles on the roads."
The Safety Pool™ Mission
Right now, there’s a lot of speculation around cars with automated driving and assistance features; many lack faith in AI and technology and would rather maintain full control of a vehicle themselves. As a result, the Safety Pool™ intends to put peoples minds at rest. With the support of The World Economic Forum, the initiative is simple in theory yet difficult in practice: to unite the growing autonomous vehicle community to make ADSs safer. The movement will look to build new frameworks, infrastructures, and processes that will be shared with industry leaders, academics, and policymakers worldwide to ensure that, going forward, autonomous vehicles systems undergo transparent testing, validation, and certification.
According to Siddartha Khastgir, from WMG, University of Warwick: "Safety of automated driving systems is a hard research challenge and can only be solved by national and international collaboration and knowledge sharing. With the launch of Safety Pool™, we are inching closer to seeing automated driving systems on the roads. Testing and validating automated driving systems transparently in an integrated simulation-based framework and in real-world scenarios will not only provide insights into the readiness of ADS but also speed up the adoption globally. We are excited to be at the forefront of this revolution."
Facebook Develops AI to Crackdown on Deepfakes
In light of the large tidal wave of increasingly believable deepfake images and videos that have been hitting the feeds of every major social media and news outlet in recent years, global organisations have started to consider the risk factor behind them. While the majority of deepfakes are created purely for amusement, their increasing sophistication is leading to a very simple question: What happens when a deepfake is produced not for amusement, but for malicious intent on a grander scale?
Yesterday, Facebook revealed that it was also concerned by that very question and that it had decided to take a stand against deepfakes. In partnership with Michigan State University, the social media giant presented “a research method of detecting and attributing deepfakes that relies on reverse engineering from a single AI-generated image to the generative model used to produce it.”
The promise is that Facebook’s method will facilitate deepfake detection and tracing in real-world settings, where the deepfake image itself is often the only information detectors have to work with.
Why Reverse Engineering?
Right now, researchers identify deepfakes through two primary methods: detection, which distinguishes between real and deepfake images, and image attribution, which identifies whether the image was generated using one of the AI’s training models. But generative photo techniques have advanced in scale and sophistication over the past few years, and the old strategies are no longer sufficient.
First, there are only so many images presented in AI training. If the deepfake was generated by an unknown, alternative model, even artificial intelligence won’t be able to spot it—at least, until now. Reverse engineering, common practice in machine learning (ML), can uncover unique patterns left by the generating model, regardless of whether it was included in the AI’s training set. This helps discover coordinated deepfake attacks or other instances in which multiple deepfakes come from the same source.
How It Works
Before we could use deep learning to generate images, criminals and other ill-intentioned actors had a limited amount of options. Cameras only had so many tools at their disposal, and most researchers could easily identify certain makes and models. But deep learning has ushered in an age of endless options, and as a result, it’s grown increasingly difficult to identify deepfakes.
To counteract this, Facebook ran deepfakes through a fingerprint estimation network (FEN) to estimate some of their details. Fingerprints are essentially patterns left on an image due to manufacturing imperfections, and they help identify where the image came from. By evaluating the fingerprint magnitude, repetition frequency, and symmetrical frequency, Facebook then applied those constraints to predict the model’s hyperparameters.
What are hyperparameters? If you imagine a generative model as a car, hyperparameters are similar to the engine components: certain properties that distinguish your fancy automobile from others on the market. ‘Our reverse engineering technique is somewhat like recognising [the engine] components of a car based on how it sounds’, Facebook explained, ‘even if this is a new car we’ve never heard of before’.
What Did They Find?
‘On standard benchmarks, we get state-of-the-art results’, said Facebook research lead Tal Hassner. Facebook added that the fingerprint estimation network (FEN) method can be used for not only model parsing, but detection and image attribution. While this research is the first of its kind, making it difficult to assess the results, the future looks promising.
Facebook’s AI will introduce model parsing for real-world applications, increasing our understanding of deepfake detection. As cybersecurity attacks proliferate, and generative AI falls into the hands of those who would do us harm, this method could help the ‘good guys’ stay one step ahead. As Hassner explained: ‘This is a cat-and-mouse game, and it continues to be a cat-and-mouse game’.