3D integrated circuits are a key part of improving the efficiency of electronics to meet the considerable demands of consumers. They are constantly being developed, but translating theoretical findings into actual devices is not easy. Now, a new design by a research team from Japan can turn these theories into reality.

In a study recently published for the VLSI Symposium 2023, researchers from Institute of Industrial Science, The University of Tokyo have reported a deposition process for nanosheet oxide semiconductor. The oxide semiconductor resulting from this process has high carrier mobility and reliability in transistors.

3D integrated circuits are made up of multiple layers that each play a role in the overall function. Oxide semiconductors are attracting a lot of attention as materials for various circuit components because they can be processed at low temperature, while still having high carrier mobility and low charge leakage, and are able to withstand high voltages.

There are also advantages to using oxides rather than metals in processes where electrodes may be exposed to oxygen during the integration process and become oxidized.

However, developing the processes needed to reliably deposit very thin layers of oxide semiconductor materials in the manufacture of devices is challenging and has not been fully established to date. Recently, the researchers have reported an atomic layer deposition (ALD) technique that produces layers appropriate for large-scale integration.

“Using our process, we carried out a systematic study of field effect transistors (FETs) to establish their limitations and optimize their properties,” explains lead author of the study, Kaito Hikake. FETs control the current flow in a semiconductor. “We tuned the ratio of the components and adjusted the preparation conditions and our findings led to the development of a multi-gate nanosheet FET for normally-off operation and high reliability.”

The findings revealed that a FET made from the chosen oxide semiconductor by ALD had the best performance. The multi-gate nanosheet FET is believed to be the first to combine high carrier mobility and reliability characteristics with normally-off operation.

“In rapidly moving areas such as electronics, it is important to translate proof of concept findings into industrially relevant processes,” says Masaharu Kobayashi, senior author. “We believe that our study provides a robust technique that can be used to produce devices that meet the market’s need for manufacturable 3D integrated circuits with high function.”

The findings in this study could provide a solution to one of the big obstacles in the manufacturing of electronic devices with semiconductors. Hopefully, this will bring more designs of electronics with high functionality to actual products.

In the past, it was a challenge to make machines play music on the grid. Today, the challenge is the opposite.

For hundreds of years, machines have been making music. In Belgium and the Netherlands, you can still hear music from carillons in ancient bell towers. These carillons share a common feature with your cell phone: They can play music programmed by humans, and they can play by themselves.

“We easily think that digital technology is a revolution and that everything is new. However, humans have always been interested in how technologies can help them control the dimension of time in music,” says Bjørnar Sandvik, music researcher at RITMO Centre for Interdisciplinary Studies in Rhythm, Time and Motion, University of Oslo.

The practice of ‘time tinkering’

In a Ph.D. he has studied various practices of what he calls “time tinkering”—the deliberate experimentation with time, structure, and rhythm in contexts of producing machine rhythm.

“My approach is based on a simple yet often overlooked premise: The very concept of machine rhythm presupposes a process where music is stored or represented on a physical material, and thus ‘frozen’ in time. While time flows irreversibly, media technologies make it possible to manipulate and experiment with the placement of sounds along the time axis,” says Sandvik.

Humans have done this since antiquity, he adds. Grids have been a necessity.

“For example, to program self-playing carillons or music boxes, you must position small pins on a rotating cylinder. If you want it to produce rhythmic music, these pins need to be spaced at the exact right distance from each other, and for this, you need a grid—a spatialized time axis—to attach the events to.”

Even today we use the same principle, he points out.

“Many of the techniques used to compose and edit music with digital production tools are possible because the music is presented visually and graphically to us on our screens. This gives us events to move along the time axis or organize in a grid on the screen.”

An ideal that changed

In his thesis, he had a historical approach but he also investigated today’s time tinkering in machine music. The grid has influenced the way rhythms have been programmed and understood in different technological eras, he explains.

“In the past, a common challenge was to make machines play music that had already been composed as notes on paper. Thus, the ideal was to surpass the human ability to follow the notation and play on grid. In the digital age, the challenge is often the opposite,” says Sandvik.

“It is easy to get a modern computer to play on the beat. The challenge is to make it go off grid, and to do it in a human and creative way.”

Today, software programs used in music production offer automatic time correction features. There are also several other functions that can synchronize events to the exact same timing on a common grid.

“This makes it more interesting to move and juxtapose single elements in order to create rhythmic friction.”

The mechanical sound

Humans have always experimented with different types of microrhythms by deviating from a note-based norm. During the last hundred years, this has shaped the development of rhythmic genres within popular music such as jazz, rock, blues, funk and soul, Sandvik explains.

“However, for a long time it was complicated, resource-intensive and time consuming to explore such microrhythms in machine programming. This is one of the reasons why machine rhythm has gotten a reputation of being ‘mechanical’ and ‘perfect’, even though today we no longer have such boundaries.”

A new standard

At the turn of the millennium, digital recording technology made it possible to move sounds along the time axis in a new and far more flexible way. This led to a new trend where songwriters seriously began experimenting with microrhythm in pop music.

Today, manipulation of time on a micro level is central to the composition practice of music producers—in fact, it is a new standard, Sandvik claims.

“Digital recording technology makes it much easier to try and retry different timings. You can make repeated recordings without losing the original or the sound quality. It is easy to program intricate rhythmic patterns, or place recorded sounds wherever you want along the time axis on your screen.”

It is (almost) unnoticeable

As part of the research project Timing and Sound in Musical Microrhythm (TIME), Sandvik interviewed several producers of electronic dance music (EDM) and he even analyzed their music.

“The fact that EDM is characterized as dance music may to some people seem like a paradox. According to rhythm research, music should deviate from the beat if it is to provide groove and a desire to dance. We often experience the rhythms in EDM as mechanistic and strictly on the grid,” says Sandvik.

However, according to findings from the TIME project and other research, it only takes a few milliseconds of deviation to create an experience of groove—in fact, listeners often do not even notice it. Through their study of the practices of EDM production, Sandvik and his colleagues concluded that producers take several measures to create a groove.

“Producers work hard to achieve rhythmic friction against the grid, either by moving the temporal onset of events or by shaping how the sounds and their intensity themselves unfold in time. Such techniques are crucial for the grooves to be successful.”

Block ciphers, a branch of modern cryptography, are playing a more prominent role in protecting information security as 5G technology develops. Although encryption algorithms of the traditional Feistel structure have great advantages in terms of consistent encryption and decryption, they have poor diffusion effects.

Additionally, they cannot adapt to the high throughput communication environment and resource-constrained devices. The S-box is the crucial nonlinear component in the block cipher and significantly determines the security of an algorithm. Unfortunately, the vast proportion of S-boxes exist in a static manner, which makes it difficult to effectively resist cryptographic attacks based on specific S-boxes.

To solve the problems, a research team led by Lang Li published their new research in Frontiers of Computer Science.

The team proposed a lightweight block cipher based on dynamic S-box named DBST for devices with limited hardware resources and high throughput requirements. The round function of DBST employs a novel generalized Feistel variant structure, which dramatically improves the diffusivity of the traditional Feistel structure. The S-box in the algorithm integrates bit-slice technology with subkeys to create a key-dependent dynamic S-box model that compensates for the shortcomings of static S-boxes.

In the research, they perform security analysis and hardware experiments on DBST. The experimental data demonstrate that the proposed algorithm has high security, high throughput rate and low hardware resources. Furthermore, differential analysis of the S-boxes proves that DBST’s S-boxes have fewer differential properties than RECTANGLE’s S-boxes.

Provided by
Frontiers Journals

Umeå University researchers have made a breakthrough that may make hydrogen—a clean, CO₂-free fuel—more affordable. The team has developed a new method that improves how hydrogen gas is produced from water and electricity, a process that’s crucial in our shift toward a more sustainable society.

This major advancement comes from a study led by Eduardo Gracia, a researcher at the Department of Physics at Umeå University. The findings of the study have been recently published in Communications Engineering.

Hydrogen gas is an excellent energy source that can be used to replace fossil fuels. It is produced through a process called water electrolysis where water is split into hydrogen and oxygen. The process requires an electrocatalyst to facilitate the reaction, and nowadays the most efficient technology for such a process is the proton exchange membrane (PEM) water electrolysis.

Metal dissolution: An issue that must be tackled

However, hydrogen production via PEM water electrolysis has a significant issue—it requires the use of noble metals such as platinum, ruthenium, and iridium. Although these metals are good at their job, they are not only expensive and limited in supply, but ruthenium and iridium also tend to break down over time.

“The breakdown of noble metals, a phenomenon known as ‘metal dissolution,’ reduces the effectiveness of hydrogen production. It’s a problem that needs to be solved for us to fully take advantage of PEM technology,” says Associate Professor Eduardo Gracia.

Stabilizing noble metals

So, if PEM technology is expected to drive the transition towards a sustainable society, we first need to tackle the strong electrocatalysts degradation. But how? Well by trapping the highly active but expensive metal in a stable but inactive “scaffold.”

This is where the Umeå team’s breakthrough comes in. The researchers, led by Eduardo Gracia, developed a new scaffold—a kind of supporting structure—that can keep the noble metals stable even under tough conditions.

This scaffold is made of a mixture of tin, antimony, molybdenum, and tungsten oxides (Sn-Sb-Mo-W), which proved to be strong enough to protect not only the noble metals but also other components of the system from breakdown during the process.

By ensuring the noble metals can last longer, the team’s findings can make PEM technology more affordable and effective for large-scale, renewable hydrogen production. This represents a key step in making our transition to a more sustainable society a reality.

Shares of GameStop are plunging before the opening bell after the company fired CEO Matthew Furlong, the former Amazon executive that was brought in two years ago to turn the struggling video game retailer around.

The company gave no reason for the dismissal and named Ryan Cohen, the company’s biggest investor, as executive chairman. Cohen sent a cryptic tweet that read “Not for long” around the time the company announced Furlong’s firing.

GameStop said Cohen will oversee investment and management of the company.

Shares tumbled more than 19% in premarket trading Thursday.

Furlong was named GameStop’s CEO in June 2021 with the mandate of heading the company’s digital remake. He had been the executive that oversaw Amazon’s Australia business and spent nine years with the company. Furlong was one of two Amazon executives hired at the time, the other being Mike Recupero, hired as GameStop’s chief financial officer.

Cohen’s holding company RC Ventures is the biggest investor in GameStop, holding an approximately 12% stake. Cohen co-founded Chewy, the online pet supply company, and had hoped to modernize the GameStop, founded in 1984.

Cohen began snapping up large stakes of GameStop at a time when the Grapevine, Texas, company was being buffeted by new technology. Gamers no longer needed GameStop because they were downloading games, rather than buying digital discs.

Furlong and other executives were brought it to execute Cohen’s goal of getting GameStop more online.

After building a massive stake, Cohen joined GameStop’s board in January 2021, along with two of his former Chewy colleagues.

GameStop became the embodiment of the “meme stock” craze two years ago, when a fanatical band of smaller-pocketed and novice investors encouraged each other to pile in. That helped trigger a “short squeeze,” on larger institutional Wall Street firms that were betting the company would continue to flounder.

The gambit worked and shares spiked more than 8,000% in 2021.

Shares have fallen drastically since then and now trade for around $20 each, which was about the cost of a share before the meme craze.

© 2023 The Associated Press. All rights reserved. This material may not be published, broadcast, rewritten or redistributed without permission.

Instagram is the main platform used by pedophile networks to promote and sell content showing child sexual abuse, according to a report by Stanford University and the Wall Street Journal.

“Large networks of accounts that appear to be operated by minors are openly advertising self-generated child sexual abuse material for sale,” said researchers at the US university’s Cyber Policy Center.

“Instagram is currently the most important platform for these networks with features like recommendation algorithms and direct messaging that help connect buyers and sellers.”

According to the Journal, a simple search for sexually explicit keywords specificaly referencing children leads to accounts that use these terms to advertise content showing sexual abuse of minors.

The profiles often “claim to be driven by the children themselves and use overtly sexual pseudonyms”, the article detailed.

While not specifically saying they sell these images, the accounts do feature menus with options, including in some cases specific sex acts.

Stanford researchers also spotted offers for videos with bestiality and self-harm.

“At a certain price, children are available for in-person ‘meetings’,” the article continued.

Meta, Instagram’s parent company, did not immediately respond to a request for comment from AFP.

According to the Journal, the social media giant acknowledged problems within its security services and said it had created a task force to address the issues raised.

Last March, pension and investment funds filed a complaint against Meta for having “turned a blind eye” to human trafficking and child sex abuse images on its platforms.

© 2023 AFP

You’re in an autonomous car when a rabbit suddenly hops onto the road in front of you.

Here’s what typically happens: the car’s sensors capture images of the rabbit; those images are sent to a computer where they are processed and used to make a decision; that decision is sent to the car’s controls, which are adjusted to safely avoid the rabbit. Crisis averted.

This is just one example of computer vision—a field of AI that enables computers to acquire, process, and analyze digital images, and make recommendations or decisions based on that analysis.

The computer vision market is growing rapidly, and includes everything from DoD drone surveillance, to commercially available smart glasses, to rabbit-avoiding autonomous vehicle systems. Because of this, there is increased interest in improving the technology. Researchers at USC Viterbi’s Information Sciences Institute (ISI) and the Ming Hsieh Department of Electrical and Computer Engineering (ECE) have recently completed Phases 1 and 2 of a DARPA (Defense Advanced Research Projects Agency) project looking to make advances in computer vision.

Two jobs spread over two separate platforms

In the rabbit-in-the-road scenario above, on the “front end” is the vision sensing (where the car’s sensors capture the rabbit’s image) and on the “back end” is the vision processing (where the data is analyzed). These are conducted on different platforms, which are traditionally physically separated.

Ajey Jacob, Director of Advanced Electronics at ISI explains the effect of this: “In applications requiring large amounts of data to be sent from the image sensor to the backend processors, physically separated systems and hardware lead to bottlenecks in throughput, bandwidth and energy efficiency.”

In order to avoid that bottleneck, some researchers approach the problem from a proximity standpoint—studying how to bring the backend processing closer to the frontend image collection. Jacob explained this methodology. “You can bring that processing onto a CPU [computer] and place the CPU closer to the sensor. The sensor is going to collect the information and send it to the computer. If we assume this is for a car, it’s fine. I can have a CPU in the car to do the processing. However, let’s assume I have a drone. I cannot take this computer inside the drone because the CPU is huge. Plus, I’ll need to make sure that the drone has an Internet connection and a battery large enough for this data package to be sent.”

So the ISI/ECE team took another approach, and looked at reducing or eliminating the backend processing altogether. Jacob states, “What we said is, let’s do the computation on the pixel itself. So you don’t need the computer. You don’t need to create another processing unit. You do the processing locally, on the chip.”

Front-end processing inside a pixel

Processing on the image sensor chip for AI applications is known as in-pixel intelligent processing (IP2). With IP2, the processing occurs right under the data on the pixel itself, and only relevant information is extracted. This is possible thanks to advances in computer microchips, specifically CMOS (complementary metal–oxide–semiconductors), which are used for image processing.

The team has proposed a novel IP2 paradigm called processing-in-pixel-in-memory (P2M) which leverages advanced CMOS technologies to enable the pixel array to perform a wider range of complex operations—including image processing.

Akhilesh Jaiswal, a computer scientist at ISI and assistant professor at ECE, led the front-end circuit design. He explained, “We have proposed a new way of fusing together sensing, memory and computing within a camera chip by combing, for the first time, advances in mixed signal analog computing and coupling them with strides being made in 3D integration of semiconductor chips.”

After processing in-pixel, only the compressed meaningful data is transmitted downstream to the AI processor, significantly reducing power consumption and bandwidth. “A lot of work went into figuring out the right trade-off between compression and computing on the pixel sensor,” said Joe Mathai, senior research engineer at ISI.

After analyzing that trade-off, the team created a framework that reduces the chip to the size of a sensor. And the data that is transferred from the sensor to the computer is also very small, because data is first pruned, or computed on the pixel itself.

From the front to the back and into the future

RPIXELS (Recurrent Neural Network Processing In-Pixel for Efficient Low-energy Heterogeneous Systems) is the resulting proposed solution for the DARPA challenge. It combines the front-end in-pixel processing with a back end that the ISI team has optimized to support the front.

In testing the RPIXEL framework, the team has seen promising results: a reduction in both data size and bandwidth of 13.5x (the DARPA goal was 10x reduction of both metrics).

ISI senior computer scientist Andrew Schmidt said, “RPIXELS reduces both the latency (time taken to do the image processing) and needed bandwidth by tightly coupling the fist layers of a neural network directly into the pixel for computing. This allows for faster decisions to be made based on what is ‘seen’ by the sensor. It also enables researchers to develop novel back end object detection and tracking algorithms to continue to innovate for more accurate and higher performance systems.”

“This project is a wonderful example of collaboration between the USC ECE department and ISI,” said Peter Bereel, Professor of Computer and Electrical Engineering at ECE. “We’ve mixed ECE’s expertise at the boundary between hardware and machine learning algorithms with the device, circuit and machine learning application expertise at ISI.”

The next step is to create a physical chip by putting the circuit onto a silicon and testing it in the real world, which could, among other things, save some rabbits.

Recent claims that artificial intelligence (AI) poses an existential threat to humanity seem to have jolted Prime Minister Rishi Sunak into action. Despite being seen as having a “pro-techology” stance, he appears to be quickly shifting position.

The Centre for AI Safety recently made mitigating the risk of extinction from AI a global priority. Against this background of caution, Sunak now reportedly wants the UK to lead in the development of guardrails to regulate AI growth.

During a trip to the US, Sunak was expected to try to persuade US president Joe Biden that the UK should play such a leading role on global AI guidelines, pitching the UK as the ideal hub for AI regulation. He seems to have met with limited success. So, is the case for this strong enough to persuade the US, and other global leaders?

Part of the anticipated pitch is that “the UK could promote a model of regulation that would be less ‘draconian’ than the approach taken by the EU, while more stringent than any framework in the US”. This is likely to raise some eyebrows and ruffle some feathers.

In part, this is because the UK’s “principles-based” approach can hardly be considered stringent at all. In its March 2023 white paper, the UK government laid out its “pro-innovation approach” to AI regulation. White papers are policy documents setting out plans for future legislation. The plans have been criticized for being too lax, already outdated, and lacking in meaningful detail.

Fit for export?

Even the Information Commissioner’s Office (ICO), one of the UK’s regulators affected by the white paper, was quick to point out its shortcomings. In this light, it does not seem to be a prime candidate for regulatory export.

Moreover, the US and the EU are making significant strides in coordinating their approaches to technology regulation. Only last week, they launched three joint expert groups to move forward with their December 2022 joint AI roadmap. It is unclear what the UK would bring to this table.

Finally, other major players have a much more credible track record of AI and digital regulation. The EU is close to completing the legislative process for its AI Act, initiated in 2021. This will give it a first-mover advantage in the jostle for position to advance a global standard for AI regulation.

Japan developed a principles-based approach to AI regulation back in 2019, which provides a clear alternative to the UK’s similar framework. While the international community still seems to accept that the UK could punch above its weight in tech matters, it is far from clear whether they would hand it the keys to global AI regulation.

Rishi Sunak’s bid to place the UK as a prime hub for AI regulation could also be seen as a calculated move to boost the country’s tech sector, which the Prime Minister has been bullish in promoting. This is evident from the launch in March 2023 of the Foundation Model Taskforce. With a budget of £100 million and the mission “to ensure sovereign capabilities and broad adoption of safe and reliable foundation models”, this is the PM’s push for the development of a “British ChatGPT”.

A country invested in promoting the development of “British AI”, and playing catch up with US and Chinese AI giants, could be seen as trying to secure an advantageous position in the race to AI regulation. This would help steer the development of AI global standards in ways that support the UK’s digital strategy, rather than being genuinely worried about dubious AI existential threats.

Fear of missing out?

Such “fantasy concerns” have been readily dismissed, and experts agree they have not been backed up by evidence. Experts are united on the risk of AI wiping out humanity being “close to zero” and have rejected the “doomer narratives” advanced by the tech industry. Earlier studies of AI-related existential risks have shown that they depend on human use or abuse of AI.

There has been no breakthrough to suggest any new need for regulation. The PM’s sudden change of heart could easily be read as an opportunistic intervention to reposition the UK in the global AI scene, as the current “pro-innovation” approach is clearly out of kilter.

The sincerity of the concerns behind the move is also put in question by the fact that the UK’s approach to AI regulation has consistently sidelined the importance of tackling the very real and current risks posed by AI—such as algorithmic discrimination or environmental impacts—which experts agree should be the primary focus of regulation.

Some of the ways in which the UK is seeking to generate a digital Brexit dividend pose serious threats to individual rights, such as in the data protection and digital information (No. 2) bill currently in discussion in parliament. This is at odds with a genuine will to put adequate guardrails in place to protect the public from AI-related harms.

So all in all the case looks weak. However, AI regulation will not be sorted in one go. If the UK wants to play a leading role in the future, it would do well to get its house in order. Seriously revising the March 2023 white paper and the data protection and digital information (No. 2) bill would be a good place to start.

Only by implementing effective protections and showing strong and decisive action domestically can the UK government hope to build the credibility needed to lead international efforts of AI regulation.

This article is republished from The Conversation under a Creative Commons license. Read the original article.