AI News You Missed In October

It has been an interesting month for machine learning (ML) and AI.

Last month gave us a look at the deployment and discovery of new ML models. Which can:

  • Analyse sports teams movements,
  • Improve processing and computation on microcontrollers for edge devices, and
  • Handle the backlog of images used for ocean exploration – the month of October has been momentous, to say the least.

Here is a round-up of the AI news in October.

Predicting sports teams’ movements with 80% accuracy

Researchers have discovered a fascinating new way to use predictive analysis to improve athlete performance

Algorithms developed by Cornell’s Laboratory for Intelligent Systems and Controls are able to predict volleyball actions by up to 80%. This discovery will allow teams to predict multiple actions. Including blocking, running, falling, standing and jumping, giving them the ability to prepare and strategize for specific game scenarios.

“[Silvia] Ferrari and doctoral students Junyi Dong and Qingze Huo trained the algorithms to infer hidden variables the same way humans gain their sports knowledge. By watching games. The algorithms use ML to extract data from videos of volleyball games, and then uses that data to help make predictions when shown a new set of games.”

With accuracy scores of up to 85%, the ML model looks to be a highly promising addition. To the improvement (and overall value) of not only sports games, but to self-driving cars and human-robot interactions.

Continual Learning On Smartphones and Sensors

While ML models trained on intelligent edge devices are ideal for making state-of-the-art predictions, the computation and memory requirements tend to be exorbitant at best.

To give these edge devices, well, the edge they need, the technique allows the on-device training of ML models on devices like microcontrollers. This could allow edge devices to continually learn from new data, eliminating data privacy issues, while enabling user customization.

Using what’s known as a neural network, these models use layers of node interconnections, or neurons, that process data to complete a task. An example includes image recognition or interactive chatbots.

According to ScienceDaily, “Training a machine-learning model on an intelligent edge device allows it to adapt to new data and make better predictions. For instance, training a model on a smart keyboard could enable the keyboard to continually learn from the user’s writing. However, the training process requires so much memory that it is typically done using powerful computers at a data center, before the model is deployed on a device.”

While this type of training is costly and raises privacy issues, researchers at MIT and the MIT-IBM Watson AI Lab developed a new technique “that enables on-device training using less than a quarter of a megabyte of memory.”

The new findings will help reduce the amount of computation needed to train a model, making the process faster and more memory efficient. In other words, their technique can train a ML model on a microcontroller in a matter of minutes.

Ocean Exploration Through AI: FathomNet

Machine learning (ML) and AI are killing it in the enterprise, medical, and transportation spaces, but still face challenges in ocean exploration.

Until now.

With a lack of standard images to train ML models for the recognition and cataloguing of aquatic objects and life, an open-source image database is helping process the backlog of image data.

Known as FathomNet, it uses AI and ML technologies to help “alleviate the bottleneck for analysing underwater imagery and accelerate important research around ocean health.”

According to Ben Woodward, co-founder and CEO of CVision AI and a co-founder of FathomNet, “In the past five years, ML has revolutionised the landscape of automated visual analysis, driven largely by massive collections of labelled data… With FathomNet, we aim to provide a rich, interesting benchmark to engage the machine-learning community in a new domain.”

Taking up more than 70% of our planet’s surface, FathomNet looks to help accelerate the exploration, understanding and problem-solving that’s needed in our oceans.

Looking For More Digital Transformation Content?

Then you have to check out our blog! 

It’s packed with tech news, tips and insights to take your digital enterprise to new heights. 

If you enjoy getting a brief update on AI news, or love to read all about big data, digital transformation, machine learning and so much more – then be sure to check out the rest of our content…

More in the Blog

Stay informed on all things AI...

< Get the latest AI news >

Join Our Webinar Cloud Migration with a twist

Aug 18, 2022 03:00 PM BST / 04:00 PM SAST