InSNU AIIS BlogbySNU AI·Jul 24, 2022A Systematic Methodology for Continuous-Time Analysis of Accelerated Gradient MethodsIn machine learning, we often minimize the loss function so that the difference between the solution point and the estimates converges to…
InSNU AIIS BlogbySNU AI·Jul 1, 2022Check Factual Consistency of Auto-generated Summaries Using Masked SummarizationKey information in source texts and reference summaries is masked to artificially generate factually inconsistent summaries
InSNU AIIS BlogbySNU AI·Jun 6, 2022The Intersection of Human Bias and Artificial IntelligenceUnderstanding how and why AI reflects human bias like a mirror helps to shape balanced AI regulatory governanceA response icon1A response icon1
InSNU AIIS BlogbySNU AI·Apr 3, 2022Measuring and Explaining the Inter-Cluster Reliability of Multidimensional ProjectionsDistortions occur when reducing dimensionality. Our metrics quantitatively measure how well the low-dimensional projection preserves the…
InSNU AIIS BlogbySNU AI·Apr 3, 2022Visual Graph Memory with Unsupervised Representation for Visual NavigationHow can an agent navigate through a particular environment to reach target observation, only using visual sensory signals?
InSNU AIIS BlogbySNU AI·Apr 3, 2022Puzzle Mix: Exploiting Saliency and Local Statistics for Optimal MixupPuzzle Mix is a data augmentation method useful for optimally leveraging the saliency information respecting the local statistics of the..
InSNU AIIS BlogbySNU AI·Apr 2, 2022Imperative-Symbolic Co-Execution of Imperative Deep Learning ProgramsOur approach can handle any imperative DL program while achieving the optimized performance of symbolic execution.
InSNU AIIS BlogbySNU AI·Apr 2, 2022Model-based Domain Randomization of Dynamics System with Deep Bayesian Locally Linear EmbeddingDomain randomization is a simple technique to enhance the robustness of the policy in various environments. We aim for efficient policy…A response icon1A response icon1
InSNU AIIS BlogbySNU AI·Apr 2, 2022Refurbish Your Training Data for Faster DNN TrainingData refurbishing is a novel sample reuse mechanism that accelerates DNN training while preserving model generalization.
InSNU AIIS BlogbySNU AI·Apr 1, 2022Expressive 3D Human Pose and Shape Estimation, Part 2: Mesh Estimation and 3D Rotational Pose…Here’s how we estimate the human mesh and capture all human parts, including hands and the face, for better human behavior understanding.