Trending...
- California: Three years after the end of Roe, Governor Newsom, First Partner sound the alarm on Trump's "Big, Beautiful" plan to defund Planned Parenthood - 121
- Ascent Solar Technologies Enters Collaborative Agreement Notice with NASA to Advance Development of Thin-Film PV Power Beaming Capabilities: ASTI - 104
- California awards over $15 million to apprenticeship programs connecting youth to high-paying jobs
LOS ANGELES - Californer -- Uncertainty estimation is critical to improving the reliability of deep neural networks. A research team led by Aydogan Ozcan at the University of California, Los Angeles, has introduced an uncertainty quantification method that uses cycle consistency to enhance the reliability of deep neural networks in solving inverse imaging problems.
This research was published on Dec. 21 in Intelligent Computing, a Science Partner Journal.
Deep neural networks have been used to solve inverse imaging problems, such as image denoising, super-resolution imaging and medical image reconstruction, in which the goal is to create an ideal image using the raw image data actually captured, often after some degradation. However, deep neural networks sometimes produce unreliable results, and in some contexts, incorrect predictions can have severe consequences. Models that can quantitatively estimate how certain they are about their output can perform better at detecting abnormal situations, such as anomalous data and outright attacks.
The new method for estimating network uncertainty uses a physical forward model, which serves as a computational representation of the underlying processes governing the input–output relationship. By combining this model with a neural network and executing forward–backward cycles between the input and output data, uncertainty accumulates and can be effectively estimated.
More on The Californer
The theoretical underpinning of the method involves establishing the bounds of cycle consistency, defined as the difference between adjacent outputs in the cycle. The researchers derived upper and lower bounds for cycle consistency, demonstrating its relationship with the uncertainty of the output of the neural network. The study considered cases where cycle outputs diverged and cases where they converged, providing expressions for both scenarios. The derived bounds can be used to estimate uncertainty even without knowledge of the ground truth.
The researchers anticipate that their cycle-consistency-based uncertainty quantification method will significantly contribute to enhancing the reliability of neural network inferences in inverse imaging problems. Additionally, the method could find applications in uncertainty-guided learning. This study marks a significant step toward addressing the challenges associated with uncertainty in neural network predictions, paving the way for more reliable and confident deployment of deep learning models in critical real-world applications.
Article: https://spj.science.org/doi/10.34133/icomputing.0071
This research was published on Dec. 21 in Intelligent Computing, a Science Partner Journal.
Deep neural networks have been used to solve inverse imaging problems, such as image denoising, super-resolution imaging and medical image reconstruction, in which the goal is to create an ideal image using the raw image data actually captured, often after some degradation. However, deep neural networks sometimes produce unreliable results, and in some contexts, incorrect predictions can have severe consequences. Models that can quantitatively estimate how certain they are about their output can perform better at detecting abnormal situations, such as anomalous data and outright attacks.
The new method for estimating network uncertainty uses a physical forward model, which serves as a computational representation of the underlying processes governing the input–output relationship. By combining this model with a neural network and executing forward–backward cycles between the input and output data, uncertainty accumulates and can be effectively estimated.
More on The Californer
- Online Distribution of the Summer 2025 Diamond Certified Directory Update Begins
- California: Governor Newsom proclaims Immigrant Heritage Month 2025
- California: Department of Defense agrees: it's time for Trump's militarization of Los Angeles to end
- Long Beach: City Launches Internet Service Enrollment Line
- Von Rock Law Founder Deidre Von Rock Named Super Lawyer for 2025
The theoretical underpinning of the method involves establishing the bounds of cycle consistency, defined as the difference between adjacent outputs in the cycle. The researchers derived upper and lower bounds for cycle consistency, demonstrating its relationship with the uncertainty of the output of the neural network. The study considered cases where cycle outputs diverged and cases where they converged, providing expressions for both scenarios. The derived bounds can be used to estimate uncertainty even without knowledge of the ground truth.
The researchers anticipate that their cycle-consistency-based uncertainty quantification method will significantly contribute to enhancing the reliability of neural network inferences in inverse imaging problems. Additionally, the method could find applications in uncertainty-guided learning. This study marks a significant step toward addressing the challenges associated with uncertainty in neural network predictions, paving the way for more reliable and confident deployment of deep learning models in critical real-world applications.
Article: https://spj.science.org/doi/10.34133/icomputing.0071
Source: UCLA ITA
Filed Under: Science
0 Comments
Latest on The Californer
- Montessori Stoppani Partners with Lifetime Montessori School
- Cymbiotika Celebrates 2025 Great Place To Work Certification™
- Long Beach: LA28 Announces Finalized Sailing Venue Plan for 2028 Olympic Games
- Individual Software Announces New Versions of its Four Typing Programs in 2025
- Britt Michaelian Brings Transformative Art & Wellness to The Ecology Center's Peace Dome
- California: Governor Newsom urges safety this Fourth of July after 600,000 pounds of illegal explosives seized
- AI-Based Neurotoxin Countermeasure Initiative Launched to Address Emerging National Security Needs: Renovaro, Inc. (N A S D A Q: RENB)
- The Naturist World Just Shifted — NaturismRE Ignites a Global Resurgence
- Mental and Emotional Self-Management, Practical Tools for Trauma-Informed Stress Management
- CGI+ Sells Multifamily Development Site in Los Angeles' South Bay to JPI for $40 Million
- MicroStrategy Incorporated (MSTR) Investors Who Lost Money Have Opportunity to Lead Securities Fraud Lawsuit
- Alten Construction awarded Encinal Junior and Senior High School project
- $796,000 in Q2 Revenue Marks Highest Earnings to Date on 3 Trailing Quarters of Profitability in Multi-Billion Homebuilding Sector: Stock Symbol: IVDN
- Cybersecurity is THE Hot Market Sector; Revenues, Earnings & Profit matter; Only 33 Million Shares + a Huge Short Position Equal an Undervalued Stock
- Despite Global Calls for a Ban, US Child Psychiatry Pushes Electroshock for Kids
- Franco Polished Plaster Celebrates 35 Years of Bringing Walls to Life in the UK
- Spartan & Guardians Partner with Guitar Legend Buckethead to Support Global Child Rescue Efforts
- Preliminary.online Introduces Short-Term Job-Readiness Courses with Employer-Verified Certifications
- Psychologist-Turned-Hermeticist Releases Modern Guide to the Seven Hermetic Principles
- Winners Announced for Asia Pacific Business Awards 2024-2025