Four-Layer AI Shield Brings Near-Real-Time Phishing Defense to Android Phones
A new multilayered machine learning framework combines URL, message, app and network analysis with adaptive user training to detect Android ...
Tag: Reinforcement Learning
Robots Learn to See, Feel and Hear at Once as Multimodal Perception Advances
A new research collection spanning tactile arrays, sim-to-real tactile learning, LiDAR-camera calibration, structured visual reasoning and virtual social touch shows ...
Simulations Show Smarter Backup Planning Cuts Cost and Delay at the Edge
Researchers have introduced RAPBS, a reliability-aware framework that places bidirectional service function chains in mobile edge computing using backup estimation, ...
Reinforcement learning upgrade helps robots stay steady in chaotic environments
An enhanced asynchronous actor-critic framework trained in simulation and on a physical mobile robot showed faster convergence, higher throughput, and ...
AI Moves Into Vascular Surgery, But Most Models Still Await Real-World Proof
A new review of thirty studies finds artificial intelligence showing strong potential across minimally invasive vascular surgery, from risk prediction ...
AI achieves first superhuman Stratego win at a fraction of the compute cost
A new AI called Ataraxos achieved the first superhuman result in Stratego while using a fraction of the compute and ...
FA-PPO Method Helps Drones Dodge Obstacles in Cluttered Skies
A new reinforcement learning method called FA-PPO was the only tested algorithm to complete simulated UAV obstacle-avoidance tasks across three ...
Reinforcement Learning Takes the Wheel of Particle Swarm Traffic Signal Tuning
A hybrid framework that lets a Proximal Policy Optimization agent switch among six search strategies cut simulated urban vehicle waiting ...
WeanNet transfers learning through fading neural connections
WeanNet progressively withdraws a parent network’s lateral influence from a student, improving adaptation in a controlled concept-drift task while enabling ...
Machine learning optimizes silk sensors for spatial motion detection
Machine-learning optimization led to three printed silk fibroin ionic conductors that generated distinct spatial resistance patterns during cyclic joint flexion.
Simulated manufacturing controller cuts emissions but misses allowance
A simulation-only manufacturing controller reduced emissions by 24.0% versus a benchmark while achieving mean resilience of 0.875, but exceeded the ...
Research report: The successor representation in high-risk drinking and alcohol-related contexts
A decision-making study of 520 adults found no overall difference in reduced successor-representation use between high- and low-risk drinkers, while ...
YOLOv5-LiDAR basketball robot reaches 97.85% detection accuracy in evaluation
A YOLOv5-LiDAR basketball-robot system achieved 97.85% detection accuracy and high task-success rates in controlled dataset evaluations and simulations, with extreme ...
Reinforcement learning curbs drag in turbulent bluff-body wakes
A wind-tunnel reinforcement-learning system reduced drag and saved energy in a turbulent bluff-body wake, outperforming model-based baselines two to fourfold.
Driverless electric taxi fleets get smarter about uncertainty
A Chalmers University modelling study combined Bayesian forecasts with transparent optimisation to coordinate large self-driving electric taxi fleets under uncertain ...
Agentic AI framework reaches 91.4% resolution success in cloud benchmarks
A cloud-native agentic AI framework combining reinforcement learning, large-language-model artifact generation and policy-as-code controls achieved 91.4% mean resolution success across ...
CARL adds stability guarantees to reinforcement-learning robot control
A reinforcement-learning method combining contraction theory was tested in simulation and on three robot platforms, providing stability guarantees and improving ...
Pareto-guided AI reduces simulated cobot welding errors and collisions
A multi-objective reinforcement-learning system reduced cobot trajectory error, collisions, cycle time, and energy use in a controlled simulation benchmark, while ...
Cloud-edge AI sends 94% less data to generate cultural patterns losslessly
A new cloud-edge framework transmits only a losslessly compressed ControlNet condition instead of the raw reference image, cutting uplink payloads ...
One Algorithm, Four Worlds: The Cross-Domain AI Map That Could Reshape Connected Systems
A new open-access cross-domain review maps how the same AI algorithms recur across IoT security, network routing, computer vision, and ...
POPULAR NEWS
-
Surgeons Resect Transverse Colon Cancer Invading Reconstructed Jejunum Following Prior Pancreaticoduodenectomy
-
Flash Heating Transforms Decommissioned Wind Turbine Blades Into Graphene and Silicon Carbide
-
INRS Hosts First North American Pasteur Network Meeting in Québec
-
Research report: Neuroimmune buffering reserve at the neurovascular unit: a CLDN5- and microglia-centred view of recurrent peripheral inflammation