Can someone copy a black-box robot brain just by watching how it acts? This study says yes. Researchers Vahid Behzadan and William Hsu show that deep reinforcement learning (DRL) agents—used in games, robotics, and trading—can be cloned via policy imitation. An attacker who can repeatedly query an agent
Self-driving cars may face no-win crash scenarios. To build fairer AI, researchers often ask the public which option they prefer, then aggregate the answers. But your moral preferences are deeply personal—and sharing them can risk your privacy. This study introduces a first privacy-preserving, crowd-guided way to train AI for
AI
Nature builds brains from a single cell. This study shows how an artificial network can do something similar - growing itself from "noise". Inspired by the early visual system (retina to LGN), the authors propose a simple developmental algorithm that starts with one "cell" and self-organizes
In plain English ARCHANGEL is a way for national archives to prove that a video hasn’t been secretly edited—today or decades from now. * AI creates a temporal content hash: a compact fingerprint from the audio and visuals across minutes or hours. * That fingerprint changes if content is tampered
Freeway on-ramps use signals to meter cars, but most systems rely on a few roadside sensors. What if the signal could watch the road and learn in real time? Researchers propose a deep reinforcement learning (DRL) system that reads live traffic camera frames and decides when to let cars onto
What if AI made plans in plain English before acting? This research shows that giving an agent a language “brainstorm” step can make it smarter and more reliable. Instead of choosing thousands of tiny moves directly, the system first writes a short plan in natural language, then a second model
AI
What if AI could learn to translate without any bilingual textbooks? This research shows how pictures can act as a bridge between languages. When we look at the same photo, speakers of different languages describe the same objects and actions. The team uses images as “pivots” so a model can
AI
Look me in the eye, camera. Ever snapped a great photo except your eyes aren’t on the camera? This research introduces GazeCorrection, an AI that subtly redirects your gaze by re-synthesizing just the eye region—keeping your identity and expression intact. How it works: a self-supervised generative model treats
What if you and an AI could sketch together? Coevo is an online platform where humans and artificial agents co-design shapes to solve different tasks. The big idea: give people and AIs a shared set of design tools and a common “language,” so both can create, learn, and iterate in
ReinforcementLearning
Reinforcement learning (RL) wins in games and simulators—but deploying it on real products is a different story. Gabriel Dulac‑Arnold, Daniel Mankowitz, and Todd Hester outline nine must-solve challenges before RL can safely power real-world systems. * Safety & constraints: avoid harmful actions while learning. * Sample efficiency: learn from limited,
machinelearning
Training AI usually needs lots of human-labeled examples. This work shows a different path: make models learn from plenty of unlabeled data by asking them to stay consistent under strong "noise." Instead of simple tweaks (like small crops or word drops), they use powerful data augmentation: RandAugment for
AI
Shopping sites host millions of product photos. Some slip through with offensive, non-compliant logos or messages, hurting customers and brands. This paper introduces an AI system that flags those images at massive scale. What’s new: a practical blend of image classification + object detection, tuned for messy, real-world retail catalogs.