Related Experiment Videos

Adaptive test-time augmentation via KL-regularized reinforcement learning for robust visual inference

Tushar Mittal¹, Arun Kumar Dubey¹, Dharmender Saini²

¹Department of Information Technology, Bharati Vidyapeeth's College of Engineering, New Delhi, India.

Scientific Reports

|July 19, 2026

Summary

This study introduces an adaptive framework using reinforcement learning to improve deep neural network accuracy under image corruptions. The method learns sample-specific transformations, enhancing robustness and reliability without model weight updates.

Area of Science:

Computer Vision
Machine Learning
Deep Learning

Background:

Deep neural networks (DNNs) exhibit reduced accuracy with real-world image corruptions and distribution shifts.
Fixed, input-agnostic test-time augmentation (TTA) has limitations in addressing these accuracy degradations.

Purpose of the Study:

To propose an adaptive framework that learns per-sample transformations for enhanced DNN robustness.
To overcome the limitations of static TTA methods.

Main Methods:

An adaptive framework using reinforcement learning (RL) was developed.
Augmentation selection was modeled as a Markov decision process with proximal policy optimization (PPO) agents.
A composite reward function combined classifier confidence gains and a self-consistency KL-divergence penalty.

Main Results:

On clean CIFAR-10, accuracy improved from 88.5% (baseline) to 90.0%.
On CIFAR-10-C, pooled top-1 accuracy increased from 75.7% (baseline) to 76.4%, outperforming static TTA and a TENT baseline.
Consistent accuracy gains were observed across various corruption types (noise, blur, weather, compression).

Conclusions:

Learned, per-sample augmentation policies significantly improve the robustness and reliability of deep vision models.
The adaptive framework operates effectively in a label-free regime, without updating model weights.
The approach demonstrates superior performance on corruptions where input-space transforms are most beneficial.

Keywords:

Adaptive image augmentation Distribution shift robustness KL-regularized policy optimization Reinforcement learning Test-time augmentation (TTA)

Related Concept Videos

Observational Learning

Observational Learning

Albert Bandura's observational learning, also known as imitation or modeling, occurs when a person observes and imitates another's behavior. It is a quicker process than operant conditioning. A well-known example is the Bobo doll study, where children who saw an adult acting aggressively towards the doll were more likely to act aggressively when left alone, compared to those who observed a nonaggressive adult. Many psychologists view observational learning as a form of latent learning because...

Articles linked to this work by shared authors, journal, and citation graph.

COVLIAS 3.5: integration of attention-based segmentation technique with fuzzy dilated convolutional neural networks for improved classification of chest X-ray scans for multiclass pneumonia diagnosis.

Scientific reports·2026

CCEO-DCABNet: Chronological Chaotic Evolution Optimization-Enabled Hybrid Deep Learning for Multiclass Disease Classification Using Chest X-Ray Images in Federated Learning.

Diagnostics (Basel, Switzerland)·2026

Geneticization studies as a theoretical field: an uneven history and conceptual reconstruction.

History and philosophy of the life sciences·2026

An intelligent gradient-guided hybrid inpainting framework for brain MRI reconstruction and Alzheimer's disease classification in connected healthcare systems.

Frontiers in medicine·2026

Electrocatalytic Water Oxidation by a Heterogenized Co (III)-Aminophenol Complex: Structure-Activity Relationship and DFT Mechanistic Insights.

Chemistry, an Asian journal·2026

Rehab-DRLX: explainable neurorehabilitation prognosis using deep reinforcement learning and transformer-based models.

Frontiers in computational neuroscience·2026

See all related articles