Category Technology/Electronics

When it comes to AI, can we Ditch the Datasets?

MIT researchers have demonstrated the use of a generative machine-learning model to create synthetic data, based on real data, that can be used to train another model for image classification. This image shows examples of the generative model’s transformation methods. Credit: Massachusetts Institute of Technology

Huge amounts of data are needed to train machine-learning models to perform image classification tasks, such as identifying damage in satellite photos following a natural disaster. However, these data are not always easy to come by. Datasets may cost millions of dollars to generate, if usable data exist in the first place, and even the best datasets often contain biases that negatively impact a model’s performance.

To circumvent some of the problems presented by datasets, M...

Read More

Magnetism helps Electrons Vanish in High-temp Superconductors

The Fermi surface on the left shows the arrangement of electrons in a copper-oxide high temperature superconductor before the “critical point.” After the critical point, the Fermi surface on the right shows that most electrons vanish. Research by the Brad Ramshaw’s lab connects this disappearance with magnetism.

A physicist’s discovery could lead to the engineering of high-temp superconducting properties into materials useful for quantum computing, medical imaging.

Superconductors — metals in which electricity flows without resistance — hold promise as the defining material of the near future, according to physicist Brad Ramshaw, and are already used in medical imaging machines, drug discovery research and quantum computers being built by Google and IBM.

However, the super-...

Read More

Engineering 2D Semiconductors with Built-in Memory Functions

A team of researchers at The University of Manchester’s National Graphene Institute (NGI) and the National Physical Laboratory (NPL) has demonstrated that slightly twisted 2D transition metal dichalcogenides (TMDs) display room-temperature ferroelectricity.

This characteristic, combined with TMDs’ outstanding optical properties, can be used to build multi-functional optoelectronic devices such as transistors and LEDs with built-in memory functions on nanometre length scale.

Ferroelectrics are materials with two or more electrically polarisable states that can be reversibly switched with the application of an external electric field. This material property is ideal for applications such as non-volatile memory, microwave devices, sensors and transistors...

Read More

Tiny ‘Skyscrapers’ help Bacteria Convert Sunlight into Electricity

Researchers have made tiny ‘skyscrapers’ for communities of bacteria, helping them to generate electricity from just sunlight and water.

The researchers, from the University of Cambridge, used 3D printing to create grids of high-rise ‘nano-housing’ where sun-loving bacteria can grow quickly. The researchers were then able to extract the bacteria’s waste electrons, left over from photosynthesis, which could be used to power small electronics.

Other research teams have extracted energy from photosynthetic bacteria, but the Cambridge researchers have found that providing them with the right kind of home increases the amount of energy they can extract by over an order of magnitude...

Read More