In this paper, we study the problem of mapping natural language instructions to complex spatial actions in a 3D blocks world. We first introduce a new dataset that pairs complex 3D spatial operations to rich natural language descriptions that require complex spatial and pragmatic interpretations such as “mirroring”, “twisting”, and “balancing”. This dataset, built on the simulation environment ...

Relational reinforcement learning combines traditional reinforcement learning with a strong emphasis on a relational (rather than attribute-value) representation. Earlier work used relational reinforcement learning on a learning version of the classic Blocks World planning problem (a version where the learner does not know what the result of taking an action will be). “Structural” learning resu...

Looking for string vacua with fixed moduli, we study compactifications of type IIA string theory on Calabi-Yau fourfolds in the presence of generic Ramond-Ramond fields. We explicitly derive the (super)potential induced by Ramond-Ramond fluxes performing a Kaluza-Klein reduction of the ten-dimensional effective action. This can be conveniently achieved in a formulation of the massive type IIA s...

We study ‘Myers effect’ for a bunch of D1-branes with IIB superstrings moving in one direction on it. We show that the ‘blown-up’ configuration is the spinning helical D1-brane. It is self-supported from collapse by the momentum flow on it. The tilting angle of the helix is determined by the number of D1-branes. The radius of the helix is stabilized to a certain value depending on the number of...

This paper considers three-dimensional (3-D) scene analysis and explores the role of models for this task. Both object and task models are included and the issues of model design and the mapping from image to model are raised. The imaging process produces a two-dimensional (2-D) image of a real-worlcl scene, while in the modelling process, a model of the real-world objects and tasks is built. T...

Most work on Predictive Representations of State (PSRs) has focused on learning and planning in unstructured domains (for example, those represented by flat POMDPs). This paper extends PSRs to represent relational knowledge about domains, so that they can use policies that generalize across different tasks, capture knowledge that ignores irrelevant attributes of objects, and represent policies ...

We construct generalized 11D supergravity solutions of fully localized intersecting D2/D4 brane systems. These solutions are obtained by embedding six-dimensional resolved Eguchi-Hanson conifolds lifted to M-theory. We reduce these solutions to ten dimensions, obtaining new D-brane systems in type IIA supergravity. We discuss the limits in which the dynamics of the D2 brane decouples from the b...

Although 1972 was the year that Winograd published his now classic natural language Study of the blocks world, that fact had not yet penetrated to the ACL. At that time people with AI computational interests were strictly in a minority in the association and it was a radical move to appoint Roger Schank as program chairman for the year's meeting. That was also the year that we didn't have a pre...

On promoting the type IIA side of the N = 1 Heterotic/type IIA dual pairs of [1] to M-theory on a 'barely G 2 Manifold' of [2], by spectrum-matching we show a possible triality between Heterotic on a self-mirror Calabi-Yau, M-theory on the above 'barely G 2-Manifold' constructed from the Calabi-Yau on the type IIA side and F-theory on an elliptically fibered Calabi-Yau 4-fold fibered over a tri...

Let n denote the lattice of partitions of an n-set, ordered by reenement. We show that for all large n there exist antichains in n whose size exceeds n 1=35 S (n; K n). Here S (n; K n) is the largest Stirling number of the second kind for xed n, which equals the largest rank within n. Some of the more complicated aspects of our previous proof of this result are avoided, and the variance of a ce...