Diffusion might be a better way to model probability in PPLs
DMVI uses diffusion models to approximate the probability distributions for faster, more accurate automated inference.
Mike Young
FACTOR: Detecting Deepfakes without Seeing Any Examples
As deepfake technology advances rapidly, we need better ways to detect them. Most current methods train on available deepfakes, but fail to generalize to new, unseen ones.
Anomaly Detection in Multivariate Time Series with... Diffusion Models?
Diffusion processes are great at smoothing out normal patterns while amplifying anomalies - perfect for AD.
Telling GPT-4 you're scared or under pressure improves performance
Researchers show LLMs respond with improved performance when prompted with emotional context
DeepMind's X-Embodiment and RT-X: Training Robots to Learn from Each Other
Just as people acquire new skills by observing others, robots also benefit from pooling knowledge
PubDef: Defending Against Transfer Attacks Using Public Models
Adversarial attacks pose a serious threat to ML models. But most proposed defenses hurt performance on clean data too much to be practical.
Swap Faces Seamlessly with the Faceswap Model
Seamlessly swap faces between pictures using AI!