artificial neural network tagged posts

Two Artificial Intelligences Talk to Each Other

Two artificial intelligences talk to each other
Tasks and models. Credit: Nature Neuroscience (2024). DOI: 10.1038/s41593-024-01607-5

Performing a new task based solely on verbal or written instructions, and then describing it to others so that they can reproduce it, is a cornerstone of human communication that still resists artificial intelligence (AI).

A team from the University of Geneva (UNIGE) has succeeded in modeling an artificial neural network capable of this cognitive prowess. After learning and performing a series of basic tasks, this AI was able to provide a linguistic description of them to a “sister” AI, which in turn performed them. These promising results, especially for robotics, are published in Nature Neuroscience.

Performing a new task without prior training, on the sole basis of verbal or written instruc...

Read More

Neuromorphic Semiconductor Device achieves World’s Highest Handwriting Pattern Recognition Rate

The world's highest level of handwriting pattern recognition rate!
A diagram of battery material-incorporated next-generation neuromorphic semiconductor device with high-density and high-reliability (left) Photo and handwriting pattern recognition accuracy of a 3-terminal-based device with high density made of a thin film of lithium-ion (right). Credit: Korea Institute of Materials Science (KIMS)

A research team led by Dr. Yong-hun Kim and Dr. Jeong-Dae Kwon has successfully developed the world’s first neuromorphic semiconductor device with high-density and high-reliability by developing a thin film of lithium-ion battery materials. They achieved this by producing ultra-thin lithium ions, a key material of lithium-ion batteries that have been in the spotlight recently, and combining it with two-dimensional nano-materials...

Read More

Can Machine-Learning Models Overcome Biased Datasets?

Caption:If researchers are training a model to classify cars in images, they want the model to learn what different cars look like. But if every Ford Thunderbird in the training dataset is shown from the front, when the trained model is given an image of a Ford Thunderbird shot from the side, it may misclassify it, even if it was trained on millions of car photos.
Credits:Image courtesy of the researchers

Researchers have applied the tools of neuroscience to study when and how an artificial neural network can overcome bias in a dataset. They found that data diversity, not dataset size, is key and that the emergence of certain types of neurons during training plays a major role in how well a neural network is able to overcome dataset bias.

Artificial intelligence systems may be able t...

Read More

AI Device Identifies Objects at the Speed of Light

The network, composed of a series of polymer layers, works using light that travels through it. Each layer is 8 centimeters square. Credit: UCLA Samueli / Ozcan Research Group

The network, composed of a series of polymer layers, works using light that travels through it. Each layer is 8 centimeters square. Credit: UCLA Samueli / Ozcan Research Group

The 3D-printed artificial neural network can be used in medicine, robotics and security. Electrical and computer engineers have created a physical artificial neural network that can analyze large volumes of data and identify objects at the actual speed of light. The device was created using a 3D printer.

Numerous devices in everyday life today use computerized cameras to identify objects – think of automated teller machines that can “read” handwritten dollar amounts when you deposit a check, or internet search engines that can quickly match photos to other similar images in their databases...

Read More