We’ve gone through yet another month where AI blows our minds.
For the month of June, we’re exploring:
- Advances in computer vision with physics
- Enhancing teamwork and collaboration with AI
- Taking on noisy DNA analysis
Get your coffee ready and dive straight into some of the amazing advancements in the AI space
Enhancing AI Computer Vision with Physics-Based Awareness
What happens when researchers try to combine physics and data to computer vision?
Researchers have proposed a new hybrid methodology to augment AI-powered computer vision systems by integrating physics-based awareness into data-driven techniques.
The study outlines three approaches to combine physics and data in computer vision AI. Which leads to improved capabilities.
“Physics-aware forms of inference can enable cars to drive more safely or surgical robots to be more precise.” Says Achuta Kadambi, Assistant Professor of Electrical and Computer Engineering at UCLA Samueli School of Engineering.
By incorporating physics into AI data sets, network architectures, and network loss functions, the researchers have achieved:
- More precise object tracking
- Accurate image generation in adverse weather conditions
- Potential for deep learning-based AIs to learn the laws of physics independently.
“With continued progress in this dual modality approach, deep learning-based AIs may even begin to learn the laws of physics on their own.” – Research Team, UCLA and US Army Research Laboratory
Enhancing Team Training Technologies With New AI Frameworks
Researchers have developed an advanced AI framework that trumps previous technologies in analysing and categorising dialogue between individuals.
The aim? To enhance team training technologies.
The framework allows training technologies to better comprehend the coordination and collaboration among team members.
“We’ve developed a new framework that significantly improves the ability of AI to analyse communication between team members. This is a significant step forward for the development of adaptive training technologies that aim to facilitate effective team communication and collaboration.” Said Jay Pande, Ph.D. Student, North Carolina State University
By leveraging the Text-to-Text Transfer Transformer (T5) model, which was trained on a large language dataset and customised using data from the U.S. Army training exercises, the AI framework incorporates contextual features of team communication.
Testing the framework’s performance against two previous AI technologies demonstrated its superior ability to classify dialogue and track the flow of information
within a squad.
“The boost in performance over the previous AI models was notable—even though we were testing the model in a new set of circumstances,” says Wookhee Min, Research Scientist, North Carolina State University
The researchers also achieved impressive results using a compact version of the T5 model, allowing for fast analysis without the need for a supercomputer.
Tackling Noisy DNA Analysis With Computational Correction
AI predictions surrounding DNA analysis tend to be noisy.
Assistant Professor Peter Koo from Cold Spring Harbor Laboratory has discovered that scientists using popular computational tools to interpret AI predictions in DNA analysis are encountering excessive “noise” or extraneous information.
To tackle this issue, Koo and his team have developed a solution that involves adding a few lines of code to obtain more reliable explanations from deep neural networks.
“The deep neural network is incorporating this random behaviour because it learns a function everywhere. But DNA is only in a small subspace of that. And it introduces a lot of noise. And so we show that this problem actually does introduce a lot of noise across a wide variety of prominent AI models.” – Peter Koo, Assistant Professor, Cold Spring Harbor Laboratory
By eliminating the noise, scientists can better identify significant DNA features, potentially leading to breakthroughs in health and medicine.
The noise arises from a lack of critical information in the data used to train AI models. This results in blind spots. Koo’s computational correction improves the accuracy of DNA analysis conducted by AI, eliminating spurious noise and enhancing the clarity of important nucleotide sites.
According to Koo, “We end up seeing sites that become much more crisp and clean, and there is less spurious noise in other regions. One-off nucleotides that are deemed to be very important all of a sudden disappear.”
Get Your Fix of AI Content
If you’re a fan of AI, machine learning, cloud and other disruptive technologies, then we’ve got some great news for you.
As an AI-enablement professional services business, we understand the complexity behind the technologies that drive the AI value chain. Whether its cloud, data, or ML engineering, we’ve got you covered!
Check out our blog for more captivating tech content.