Failure case 2: Augerino depends on the used parameterisation of invariance. The full GGN approximation in Eq. 5 is inO(NP2C) for computingN matrix-products. The diagonalGGNapproximation would be inO(NPC)and computation of the log-determinant onlyO(P). Computing the log-determinant can be done efficiently inO(D3 +G3)by decomposing the Kronecker factors (Immer et al., 2021a). The last two terms dependent onS come up due to the aggregation ofaugmentation samples inour approximation, that is,the expectations overaandg in the second line of Eq. 15.

Can AI really help us discover new materials? Judging from headlines and social media posts in recent years, one might reasonably assume that AI is going to fix the power grid, cure the world's diseases, and finish my holiday shopping for me. This week, we published a new package called Hype Correction . The collection of stories takes a look at how the world is starting to reckon with the reality of what AI can do, and what's just fluff. One of my favorite stories in that package comes from my colleague David Rotman, who took a hard look at AI for materials research . AI could transform the process of discovering new materials--innovation that could be especially useful in the world of climate tech, which needs new batteries, semiconductors, magnets, and more.

We introduce Language-Informed Latent Actions (LILA), a framework for learning natural language interfaces in the context of human-robot collaboration. LILA falls under the shared autonomy paradigm: in addition to providing discrete language inputs, humans are given a low-dimensional controller $-$ e.g., a 2 degree-of-freedom (DoF) joystick that can move left/right and up/down $-$ for operating the robot. LILA learns to use language to modulate this controller, providing users with a language-informed control space: given an instruction like "place the cereal bowl on the tray," LILA may learn a 2-DoF space where one dimension controls the distance from the robot's end-effector to the bowl, and the other dimension controls the robot's end-effector pose relative to the grasp point on the bowl. We evaluate LILA with real-world user studies, where users can provide a language instruction while operating a 7-DoF Franka Emika Panda Arm to complete a series of complex manipulation tasks. We show that LILA models are not only more sample efficient and performant than imitation learning and end-effector control baselines, but that they are also qualitatively preferred by users.

When pet Lila wasn't getting as much playtime as the other two animals in her Plymouth, Mass., home, owner Genie Boutchia felt guilty. Then when a potential new owner came calling with $850 in hand, Ms. Boutchia felt even guiltier. She changed her mind and deemed Lila not for sale. Such feelings of moral responsibility might seem normal, even admirable, in a dog owner. But Lila is not a real dog.