Students will gain knowledge in robot decision-making by critically reviewing existing literature in this field, with a focus on semantics-informed approaches. In addition to the literature review, students are expected to implement relevant work to showcase their understanding of the approach and support the claims in their literature analysis. The specific topics covered will change each term, encompassing areas such as safe robot learning, learning from demonstration, language-conditioned robot learning, spatial AI, and 3D scene understanding.
This seminar will discuss recent developments in modern machine learning towards capturing geometric and semantic understanding of the 3D environments around us. Topics will include deep learning approaches for both generative and discriminative 3D tasks on various 3D representations such as point clouds, voxels, meshes.
Takes an academic research focus on cutting-edge topics in computer vision, graphics, and machine learning. Topics cover 3D reconstruction, 3D semantic scene understanding, generative 3D modeling, dynamic modeling, self-supervised, weakly-supervised, and few-shot learning for 3D reconstruction and semantics.
Explore state-of-the-art algorithms for both supervised and unsupervised machine learning on 3D data, for both analysis and synthesis of 3D shapes and scenes.
Students will gain knowledge in robot decision-making by critically reviewing existing literature in this field, with a focus on semantics-informed approaches. In addition to the literature review, students are expected to implement relevant work to showcase their understanding of the approach and support the claims in their literature analysis. The specific topics covered will change each term, encompassing areas such as safe robot learning, learning from demonstration, language-conditioned robot learning, spatial AI, and 3D scene understanding.
This seminar will discuss recent developments in modern machine learning towards capturing geometric and semantic understanding of the 3D environments around us. Topics will include deep learning approaches for both generative and discriminative 3D tasks on various 3D representations such as point clouds, voxels, meshes.
Takes an academic research focus on cutting-edge topics in computer vision, graphics, and machine learning. Topics cover 3D reconstruction, 3D semantic scene understanding, generative 3D modeling, dynamic modeling, self-supervised, weakly-supervised, and few-shot learning for 3D reconstruction and semantics.
Explore state-of-the-art algorithms for both supervised and unsupervised machine learning on 3D data, for both analysis and synthesis of 3D shapes and scenes.
This lecture focuses on modern machine learning techniques, such as convolutional neural networks, recurrent neural networks, and generative techniques.