Physics

Neurophysiological studies with animals suggest that sounds modulate activity in primary visual cortex in the presence of concurrent visual stimulation. Noninvasive neuroimaging studies in humans have similarly shown that sounds modulate activity in visual areas even in the absence of visual stimuli or visual task demands. However, the spatial and temporal limitations of these noninvasive methods prevent the determination of how rapidly sounds activate early visual cortex and what information about the sounds is relayed there. Using spatially and temporally precise measures of local synaptic activity acquired from depth electrodes in humans, we demonstrate that peripherally presented sounds evoke activity in the anterior portion of the contralateral, but not ipsilateral, calcarine sulcus within 28 ms of sound onset. These results suggest that auditory stimuli rapidly evoke spatially specific activity in visual cortex even in the absence of concurrent visual stimulation or visual tas...

This paper describes our entry for the INLG 2018 E2E NLG challenge. Generating fluent natural language descriptions from structured data is a key sub-task for conversational agents. In the E2E NLG challenge, the task is to generate these utterances conditioned on multiple attributes and values. Our system utilizes several extensions to the generalpurpose sequence-to-sequence (S2S) architecture to model the latent content selection process, particularly different variants of copy attention and coverage decoding. In addition, we propose a new training method based on diverse ensembling to encourage the model to learn latent plans in training. We empirically evaluate these techniques and show that the system increases the quality of generated text across five automated metrics. Out of a total of sixty submitted systems from 16 institutions, our best system ranks first-place in three of the five metrics, including ROUGE.

Being inspired by the success of \texttt{word2vec} \citep{mikolov2013distributed} in capturing analogies, we study the conjecture that analogical relations can be represented by vector spaces. Unlike many previous works that focus on the distributional semantic aspect of \texttt{word2vec}, we study the purely \emph{representational} question: can \emph{all} semantic word-word relations be represented by differences (or directions) of vectors? We call this the word2vec conjecture and point out some of its desirable implications. However, we will exhibit a class of relations that cannot be represented in this way, thus falsifying the conjecture and establishing a limitative result for the representability of semantic relations by vector spaces over fields of characteristic 0, e.g., real or complex numbers.

We propose a new complexity measure for Markov decision processes (MDPs), the maximum expected hitting cost (MEHC). This measure tightens the closely related notion of diameter [JOA10] by accounting for the reward structure. We show that this parameter replaces diameter in the upper bound on the optimal value span of an extended MDP, thus refining the associated upper bounds on the regret of several UCRL2-like algorithms. Furthermore, we show that potential-based reward shaping [NHR99] can induce equivalent reward functions with varying informativeness, as measured by MEHC. We further establish that shaping can reduce or increase MEHC by at most a factor of two in a large class of MDPs with finite MEHC and unsaturated optimal average rewards.

An important problem that attracted much recent attention is to understand a low-energy limit of a hypothetical theory that generalizes the worldvolume theory of a single M2-brane [1] to the case of N coincident M2-branes. Using some earlier ideas of [2], Bagger and Lambert, and Gustavson (BLG) succeeded in constructing a three-dimensional N ¼ 8 superconformal Chern-Simons-matter theory based on a 3-algebra [3–5].

We elaborate on the suggestion made in arXiv:0806.3498 that the 3d \N=8 superconformal SU(N) Chern-Simons-matter theory of Lorentzian Bagger-Lambert-Gustavson type (L-BLG) can be obtained by a scaling limit (involving sending the level k to infinity and redefining the fields) from the \N=6 superconformal U(N)xU(N) Chern-Simons-matter theory of Aharony, Bergman, Jafferis and Maldacena (ABJM). We show that to implement such a limit in a consistent way one is to extend the ABJM theory by an abelian "ghost" multiplet. The corresponding limit at the 3-algebra level also requires extending the non-antisymmetric Bagger-Lambert 3-algebra underlying the ABJM theory by a negative-norm generator. We draw analogy with similar scaling limits discussed previously for bosonic Chern-Simons theory and comment on some implications of this relation between the ABJM and L-BLG theories.

We elaborate on the suggestion made in arXiv:0806.3498 that the 3d \N=8 superconformal SU(N) Chern-Simons-matter theory of Lorentzian Bagger-Lambert-Gustavson type (L-BLG) can be obtained by a scaling limit (involving sending the level k to infinity and redefining the fields) from the \N=6 superconformal U(N)xU(N) Chern-Simons-matter theory of Aharony, Bergman, Jafferis and Maldacena (ABJM). We show that to implement such a limit in a consistent way one is to extend the ABJM theory by an abelian "ghost" multiplet. The corresponding limit at the 3-algebra level also requires extending the non-antisymmetric Bagger-Lambert 3-algebra underlying the ABJM theory by a negative-norm generator. We draw analogy with similar scaling limits discussed previously for bosonic Chern-Simons theory and comment on some implications of this relation between the ABJM and L-BLG theories.