Connect with us

Tech

Design principles for more reliable and trustworthy AI artists

Published

on

Design principles for more reliable and trustworthy AI artists


A conceptual illustration of theoretical analysis. Theorem 1 (Corollary 1 for diffusion model) translates the perturbation in the parameter space into the set of perturbed distributions. Theorem 2 (Corollary 2 for diffusion model) shows that flat minima lead to robustness against the distribution gap. Credit: arXiv (2025). DOI: 10.48550/arxiv.2503.11078

When users ask ChatGPT to generate an image in a Ghibli style, the actual image is created by DALL-E, a tool powered by diffusion models. Although these models produce stunning images—such as transforming photos into artistic styles, creating personalized characters, or rendering realistic landscapes—they also face certain limitations. These include occasional errors, like three-fingered hands or distorted faces, and challenges in running on devices with limited computational resources, like smartphones, due to their massive number of parameters.

A research team, jointly led by Professors Jaejun Yoo and Sung Whan Yoon of the UNIST Graduate School of Artificial Intelligence at UNIST, has proposed a new design principle for generative AI that addresses these issues. They have shown, through both and extensive experiments, that training diffusion models to reach “flat minima”—a specific type of optimal point on the loss surface—can simultaneously improve both the robustness and the generalization ability of these models.

Their study was presented at the International Conference on Computer Vision (ICCV 2025), and the findings are posted on the arXiv preprint server.

Diffusion models are widely used in popular AI applications, including tools like DALL-E and Stable Diffusion, enabling a range of tasks from style transfer and cartoon creation to realistic scene rendering. However, deploying these models often leads to challenges, such as error accumulation during short generation cycles, performance degradation after model compression techniques like quantization, and vulnerability to adversarial attacks—small, malicious input perturbations designed to deceive the models.

The research team identified that these issues stem from fundamental limitations in the models’ ability to generalize—meaning their capacity to perform reliably on new, unseen data or in unfamiliar environments.

To address this, the research team proposed guiding the training process toward “flat minima”—regions in the model’s loss landscape characterized by broad, gentle surfaces. Such minima help the model maintain stable and reliable performance despite small disturbances or noise. Conversely, “sharp minima”—narrow, steep valleys—tend to cause performance to deteriorate when faced with variations or attacks.

Among various algorithms designed to find flat minima, the team identified Sharpness-Aware Minimization (SAM) as the most effective. Models trained with SAM demonstrated reduced error accumulation during rapid generation tasks, maintained higher quality outputs after compression, and exhibited a sevenfold increase in resistance to , significantly boosting their robustness.

While previous research addressed issues like error accumulation, quantization errors, and adversarial vulnerabilities separately, this study shows that focusing on flat minima offers a unified and fundamental solution to all these challenges.

The researchers highlight that their findings go beyond simply improving . They provide a fundamental framework for designing trustworthy, versatile generative AI systems that can be effectively applied across various industries and real-world scenarios. Additionally, this approach could pave the way for training large-scale models like ChatGPT more efficiently, even with limited data.

More information:
Taehwan Lee et al, Understanding Flatness in Generative Models: Its Role and Benefits, arXiv (2025). DOI: 10.48550/arxiv.2503.11078

Journal information:
arXiv


Citation:
Design principles for more reliable and trustworthy AI artists (2025, November 6)
retrieved 6 November 2025
from https://techxplore.com/news/2025-11-principles-reliable-trustworthy-ai-artists.html

This document is subject to copyright. Apart from any fair dealing for the purpose of private study or research, no
part may be reproduced without the written permission. The content is provided for information purposes only.





Source link

Continue Reading
Click to comment

Leave a Reply

Your email address will not be published. Required fields are marked *

Tech

Top Verizon Promo Codes and Deals for November 2025

Published

on

Top Verizon Promo Codes and Deals for November 2025


Back in the day, Verizon proved their expansive reach with the spokesperson who asked “Can you hear me now?” With the spokesman pulling a Benedict Arnold and moving to T-Mobile, Verizon is now called the “Network America Relies On,” proven by their 4G LTE network covering 99% of the U.S. population, and 5G network expanding daily (though hovering around 13% now). The telecommunications conglomerate is the second-largest wireless carrier in the US and the largest network in America; now including everything from wireless home internet to headphones, smartwatches and cell phone accessories. We know how important it is to stay in communication, so we’ve found some of the best ways to save money, even without a Verizon promo code.

Latest Verizon Coupons: Up to $1,100 on Phones and Plans

You can get up to $1,100 off the newest smartphone models when you trade your phone in and get a new line on the Verizon Unlimited Ultimate plan. The Verizon promo credit will be applied over 36 months and the trade in phone must be from Apple, Google, or Samsung. With this deal, you can get a free Galaxy S25, Watch, and Tablet when you sign up for a new line on MyPlan; a free Google Pixel 10, Watch, and Tablet with a new MyPlan, with no trade-in needed; a free iPhone 17 Pro with a new line of Unlimited Ultimate; or a free Galaxy S25+ with a new line of Unlimited Ultimate, no trade-in required. Plus, you can get an iPhone 17 Pro Max for less than $5 per month with the Unlimited Ultimate line. And both new and existing customers can get free Google Pixel 10 Pro with a trade-in in any condition. You can also save up to $1,100 off the Samsung Galaxy Z Fold7, or get a free Motorola Rzr, with no trade-in required when you get a new Unlimited Ultimate line.

Plus, now both new and existing customers have tons of different options to save. You can get up to $1,100 off on the latest phones when you buy a new plan—along with overnight shipping included for new customers signing up for a phone plan. Plans are locked in at $0 per month for the first three years. You can also receive a free phone when you trade-in your old phone—no matter the condition.

There are also sweet deals, like $1,100 off the new Samsung Galaxy Z Fold7. You’ll get a free Samsung Galaxy S25+ when you set up a new line with an Unlimited Ultimate plan, and you’ll have a $0 monthly payment plan for 36 months.

Verizon Promo Codes and Free Phone Deals

There are even more ways to save with rotating Verizon deals. Right now, you can get cutting edge smartphones including the Galaxy Z Flip7, Fold7, iPhone 16 Pro, and Samsung Galaxy S25+ with AI for free—with no trade-in required and free 2-day shipping. More details can be found at the link, but you’ll need to get a new line on an Unlimited Plan and overall you’ll be saving $999.99 after credits of $27.77 per month for 36 months.

Verizon just seems to be handing out phones at this point—you can get a Samsung Galaxy Watch8 for free when you buy an Android and new plan when you purchase a 5G smartphone with select Unlimited Plans. You can also get a free iPhone 16 Pro, or an Apple Watch SE or iPad 10th generation for free when you purchase a 5G iPhone with select Unlimited Plans. Even without a trade-in, customers can get four iPhone 15 Plus phones when they buy a new myPlan line. Existing customers will save $830 when they buy an iPhone 15 Plus device and upgrade to (or already have) an Unlimited Plus or Unlimited Ultimate plan.

Our Favorite Verizon Phones From $5 per Month at Verizon

Apple’s iPhone 16 is available right now, and looking better than ever—it’s ready for Apple’s new AI rollout, a battery that’s easier to replace, and has a host of games. We put together a handy guide to which iPhone 16 or Galaxy you should buy, in case the various iPhone 16 models to choose from are a bit overwhelming, along with some great cases and accessories to keep your most-used device safe and stylish. If you’ve been eyeing an iPhone 16 Pro, Verizon is offering a great deal where you can get the phone for as little as $5 per month—or it’s free when you add a new line. Stay tuned here for updates, because most Verizon promo codes are for new customers.

We at WIRED review and write about the newest line of Samsung products as well, and have created a guide to help you decide which Samsung Galaxy S24 model you should buy and some of our favorite Samsung Galaxy S24 accessories to upgrade your device. Like competitors, Samsung has worked to continuously integrate AI to make the phone even smarter, helping test for Google’s Gemini.

Another popular non-iPhone model line is the Google Pixel, which are our favorite Android phones. We have a bunch of recommendations for great cases and accessories to keep your phone tip-top, and write often about the company’s ever-evolving AI features.

Verizon Home Internet Deals

Verizon also offers excellent home internet, so if you’re like me and are perpetually unhappy with your wifi speeds while WFH, now might be the perfect time to switch. Plans start at just $35 a month with a price guarantee, and $0 due when you switch. You can also save on entertainment, like $12 per month off the Disney Bundle and $7 a month off Netflix and Max.

Score up to $300 Off With Verizon Bundle Offers

Verizon has tons of ways to bundle your plans with various entertainment and streaming services to spend less. Both new and existing customers can save up to $300 a year with discounts on internet and mobile bundle deals. If you’re new to Verizon, make sure you buy the mobile plan first, then add home internet and your discount will be applied to your bill. Depending on your needs, they have add-on bundles with YouTube TV, NFL Sunday Ticket for the sports lovers, Netflix and Max with ads, and Unlimited Plus with the Disney Bundle, which includes Disney, Hulu, and ESPN+.

Students, Teachers, Military, and Nurses Can Save Up to $45 on Verizon

They also have discounts on mobile plans and home internet for students, military, teachers, and nurses. This means right now, students can get up to $12 off per month of MyPlan line and $20 off Fios Home Internet; members of the military can get discounted lines of $25 per month with 4 lines on Unlimited Welcome and $30 for Fios Home Internet, which includes an extra $15 per month when you bundle mobile with home internet; and nurses get $25 or more off per month, per line and Fios Home Internet starting at $45.

Check out our roundup of deals above, with no Verizon promo code needed to save big on your (arguably) most important tech purchases.



Source link

Continue Reading

Tech

Top Surfshark Promo Codes for November 2025

Published

on

Top Surfshark Promo Codes for November 2025


Surfshark is one of our favorite VPNs to get around geographical restrictions on content (aka access Netflix) and protect your web traffic while using an open Wi-Fi hotspot. Surfshark is secure, and it provides great value for the money if you pay for two years upfront. We also like that Surfshark’s plans are simple: A single Surfshark subscription works simultaneously on multiple devices and protects your entire household. No multiple accounts, no confusion. Get award-winning security that’s even more budget-friendly with our roundup of top Surfshark coupons, plans from $3 per month, and 3 free months on 1 or 2 year subscriptions.

Black Friday Surfshark Offers 2025

Ahead of the holiday season, Surfshark has already released its Black Friday and Cyber Monday deals, with discounts on its 24-, 12-, and one-month plans, including Starter, One, and One+ plans. The Starter plan includes a 30-day money-back guarantee and gets you a secure, fast VPN and email masking; plus, with this discount you get three extra months for free. The One plan also includes antivirus protection, dark web monitoring, and private search results; plus the three free months and a money-back guarantee. The top tier One+ plan includes all of that plus data removal by Incogni, three free months, and a money-back guarantee, starting at only $4 per month.

Save on 1 and 2 Year Plans With Surfshark Coupons

Right now, you can get a 7 day free trial, an extra 3 months for free on all 1-year and 2-year plans, and up to 87% off with active Surfshark coupon codes. The Surfshark Starter Plan is still the cheapest plan—now it starts at $2 per month on a 2-year plan compared to $20 per month if you were paying month to month. The most comprehensive plan is Surfshark One+, which is now 81% off, at $4 per month. Even the basic Surfshark Starter plan includes unlimited devices, ad and cookie pop-up blockers, double encryption, and alternative IDs.

Surfshark also offers Surfshark One Plus for $6 per month, with 3 months extra. Surfshark One adds some extra features like Real-time credit card and ID breach alerts, personal data security reports, and anti-virus and anti-malware software. One Plus gets you all that and adds data removal services. You can see the full breakdown of services and a comparison of the plans on the Surfshark website.

In my testing over the years, Surfshark has consistently had some of the best speeds of any VPN I’ve used. I have never had any problem streaming HD content through Surfshark. It’s fast enough that you won’t notice any speed degradation.

Score a 7-Day Free Trial on Surfshark and 30 Day Guarantee

If you’re feeling commitment-shy, fear not, there is a 7-day free trial on all plans, plus a 30-day refund period and if it turns out Surfshark isn’t right for you. Just contact Surfshark and you’ll receive a full refund within 30 days with their money-back guarantee. With your Surfshark free trial, you can do things like change your IP with 3,200+ encrypted servers, encrypt your connection to stop data tracking, and more.

Students Can Get a 15% Off Surfshark Discount Code

Students can get an even better deal on Surfshark, adding an additional 15% discount on VPN and Antivirus plans on top of the coupons above. To receive your discount code, first verify your student status through StudentBeans on Surfshark’s website. Graduate students can get the graduate discount of 15% off the already great price of $1.99 per month.

Get Paid $38 Or 3 Months Of SurfShark Free When You Refer A Friend

When you invite friends to use Surfshark, you can score cash rewards. With the refer a friend program, you can now choose between getting up to 3 free months of Surfshark, or paying the equivalent amount via PayPal (up to $38). If your invited friend purchases 1 month of Surfshark, you’ll both get 1 month free. If your invited friend purchases a longer Surfshark plan, you’ll both get 3 months free. To redeem, you’ll need to share the unique referral link that’ll be in your Surfshark app with as many friends as you want. When each friend signs up and keeps their account active for 31 days, you’ll be able to choose whether you want money in your account or free months added to your Surfshark subscription.



Source link

Continue Reading

Tech

Dual-level engineering strategy shows promise for high-performance lithium–sulfur batteries

Published

on

Dual-level engineering strategy shows promise for high-performance lithium–sulfur batteries


Researchers develop a novel porous carbon nanofiber interlayer material with unsaturated cobalt single-atom catalysts. Credit: Seung-Keun Park and Inho Nam from Chung-Ang University Image source link: https://link.springer.com/article/10.1007/s42765-025-00614-w

Carbon-supported single-atom catalysts with metal-N moieties are highly promising for lithium–sulfur batteries. They can enhance redox kinetics and suppress the dissolution of lithium polysulfides. However, carbon substrate structure optimization and catalyst coordination environment modulation must be done simultaneously to maximize the potential of these catalysts.

Taking on this challenge, a team of researchers led by two associate professors from Chung-Ang University—Seung-Keun Park from the Department of Advanced Materials Engineering and Inho Nam from the Department of Chemical Engineering—has demonstrated dual‑level engineering of metal–organic framework (MOF)‑derived hierarchical porous carbon nanofibers with low‑coordinated cobalt single‑atom catalysts for high‑performance lithium–sulfur batteries. Their novel findings were published in Advanced Fiber Materials on 24 September 2025.

Dr. Park says, “Our motivation lies in addressing the fundamental materials challenges that have limited the development of next-generation energy storage systems. Lithium-ion batteries have been widely adopted but are approaching their intrinsic energy density limits.

“Lithium sulfur batteries offer much higher theoretical capacity and energy density, yet they are severely restricted by the polysulfide shuttle effect, slow redox kinetics, and rapid capacity fading. Our group has long been committed to overcoming these bottlenecks by combining structural engineering of carbon frameworks with atomic-level catalyst design.”

In this study, the researchers focused on embedding single cobalt atoms in a low-coordinated N3 environment within a porous carbon nanofiber network. This approach enhances the adsorption of lithium polysulfides and accelerates their redox reactions, thereby mitigating the shuttle effect and improving overall kinetics. Therefore, the present work supports the belief that rational materials design at both the macro and atomic levels can solve long-standing challenges.







Atomic trajectory of Co–N3 over representative segment of approximately 150 fs. Credit: Advanced Fiber Materials (2025). DOI: 10.1007/s42765-025-00614-w

From a materials perspective, the proposed dual-level engineering strategy integrates a hierarchical porous carbon nanofiber structure with atomically dispersed cobalt single-atom sites in a low-coordinated N3 configuration. The carbon nanofiber provides , abundant pore channels, and excellent electrolyte wettability, while the cobalt sites catalyze the adsorption and conversion of polysulfides. This synergistic design allows the battery to achieve high-capacity retention and superior rate performance over hundreds of cycles.

In the long term, the results of this study could contribute to the realization of high-performance lithium sulfur batteries for diverse real-life applications. These include with extended driving ranges, large-scale renewable energy storage systems that can balance intermittent solar and , and lightweight, flexible power sources for portable and wearable electronics.

“Our material is free standing, binder free, and flexible. It can be directly applied as an interlayer in pouch cells and has been demonstrated to maintain mechanical integrity even under bending, while powering small devices,” points out Dr. Nam, highlighting the immense practical implications of their work.

For society, such advances mean safer and more efficient batteries that accelerate the transition to clean energy. This can reduce dependence on critical raw materials, lower costs, decrease carbon emissions, and ultimately make sustainable technologies more reliable and accessible in everyday life.

More information:
Jeong Ho Na et al, Dual-Level Engineering of MOF-Derived Hierarchical Porous Carbon Nanofibers with Low-Coordinated Cobalt Single-Atom Catalysts for High-Performance Lithium–Sulfur Batteries, Advanced Fiber Materials (2025). DOI: 10.1007/s42765-025-00614-w

Citation:
Dual-level engineering strategy shows promise for high-performance lithium–sulfur batteries (2025, November 6)
retrieved 6 November 2025
from https://techxplore.com/news/2025-11-dual-strategy-high-lithiumsulfur-batteries.html

This document is subject to copyright. Apart from any fair dealing for the purpose of private study or research, no
part may be reproduced without the written permission. The content is provided for information purposes only.





Source link

Continue Reading

Trending