Search

The success of scene graphs for visual scene understanding has brought attention to the benefits of abstracting a visual input (e.g., image) into a structured representation, where entities (people and objects) are nodes connected by edges specifying their relations. Building these representations, however, requires expensive manual annotation in the form of images paired with their scene graphs or frames. These formalisms remain limited in the nature of entities and relations they can capture. In this paper, we propose to leverage a widely-used meaning representation in the field of natural language processing, the Abstract Meaning Representation (AMR), to address these shortcomings. Compared to scene graphs, which largely emphasize spatial relationships, our visual AMR graphs are more linguistically informed, with a focus on higher-level semantic concepts extrapolated from visual input. Moreover, they allow us to generate meta-AMR graphs to unify information contained in multiple image descriptions under one representation. Through extensive experimentation and analysis, we demonstrate that we can re-purpose an existing text-to-AMR parser to parse images into AMRs. Our findings point to important future research directions for improved scene understanding., Comment: published in CoNLL 2022

Given an $m$-tempered strongly continuous action $\alpha$ of $\R$ by continuous $^{*}$-automorphisms of a Frechet $^{*}$-algebra $A$, it is shown that the enveloping \hbox{$\sigma$-$C^{*}$-algebra} $E(S(\R,A^{\infty},\alpha))$ of the smooth Schwartz crossed product $S(\R,A^{\infty},\alpha)$ of the Frechet algebra $A^{\infty}$ of $C^{\infty}$-elements of $A$ is isomorphic to the \hbox{$\sigma$-$C^{*}$-crossed} product $C^{*}(\R,E(A),\alpha)$ of the enveloping $\sigma$-$C^{*}$-algebra $E(A)$ of $A$ by the induced action. When $A$ is a hermitian $Q$-algebra, one gets $K$-theory isomorphism $RK_{*}(S(\R,A^{\infty},\alpha)) = K_{*}(C^{*}(\R,E(A),\alpha)$ for the representable $K$-theory of Frechet algebras. An application to the differential structure of a $C^{*}$-algebra defined by densely defined differential seminorms is given., Comment: 13 pages, 3 figures

For a compactly generated LCA group $G$, it is shown that the set $H(G)$ of all generalized characters on $G$ equipped with the compact-open topology is a LCA group and $H(G) = \dg$ (the dual group of $G$) if and only if $G$ is compact. Both results fail for arbitrary LCA groups. Further, if $G$ is second countable, then the Gel'fand space of the commutative convolution algebra $\ccg$ equipped with the inductive limit topology is topologically homeomorphic to $\hg$., Comment: 8 pages

Let $f$ be a continuous function on the unit circle $\Gamma$, whose Fourier series is $\omega$-absolutely convergent for some weight $\omega$ on the set of integers $\mathcal{Z}$. If $f$ is nowhere vanishing on $\Gamma$, then there exists a weight $\nu$ on $\mathcal{Z}$ such that $1/f$ had $\nu$-absolutely convergent Fourier series. This includes Wiener's classical theorem. As a corollary, it follows that if $\phi$ is holomorphic on a neighbourhood of the range of $f$, then there exists a weight $\chi$ on $\mathcal{Z}$ such that \hbox{$\phi\circ f$} has $\chi$-absolutely convergent Fourier series. This is a weighted analogue of L\'{e}vy's generalization of Wiener's theorem. In the theorems, $\nu$ and $\chi$ are non-constant if and only if $\omega$ is non-constant. In general, the results fail if $\nu$ or $\chi$ is required to be the same weight $\omega$., Comment: 4 pages

We present new HST/WFPC2 imagery and STIS long-slit spectroscopy of the planetary nebula NGC 7009. The primary goal was to obtain high spatial resolution of the intrinsic line ratio [O III] 4364/5008 and thereby evaluate the electron temperature (T_e) and the fractional mean-square T_e variation (t_A^2) across the nebula. The WFPC2 T_e map is rather uniform; almost all values are between 9000 - 11,000 K, with the higher T_e's closely coinciding with the inner He^++-zone. The results indicate very small values - <~ 0.01 - for t_A^2 throughout. Our STIS data allow an even more direct determination of T_e and t_A^2, albeit for a much smaller area than with WFPC2. We present results from binning the data along the slit into tiles that are 0.5'' square (matching the slit width). The average [O III] temperature using 45 tiles (excluding the central star and STIS fiducial bars) is 10,139 K; t_A^2 is 0.0035. The measurements of T_e reported here are an average along each line of sight. Therefore, despite finding remarkably low t_A^2, we cannot completely rule out temperature fluctuations along the line of sight as the cause of the large abundance discrepancy between heavy element abundances inferred from collisionally excited emission lines compared to those derived from recombination lines., Comment: MNRAS accepted: 26 pages, 4 Figures, 1 Table

The domain of second language reading is a rich source for insights into bilingual cognitive processing. Most cross-language transfer research has been carried out on acquisition and production of second-language (L2) structures. Relatively little work has been done on cross-language transfer in bilingual reading. An overreliance on top-down, psycholinguistic-guessing-game models in the fields of L2 reading is one reason why cross-language transfer in initial word-recognition stages of reading has not been studied vigorously. Researchers have developed several models of the reading process (particularly orthographic and phonological processing) in which cross-language transfer effects may be sought. Another variable that may affect cross-language transfer is the morphemic complexity of the words. One of the most vigorously investigated areas is how bilinguals represent the meaning of words in their two languages. Some of the richest data in L2 acquisition come from studies on syntactic transfer. Cross-language transfer of background knowledge and cultural schemata is of major concern for L2 pedagogy. As reading progresses, self-monitoring of text understanding becomes essential. More research on the effects of first-language (L1) on L2 reading is needed especially in the initial word recognition stages. Isolating the components of the reading process and investigating the nature of cross-language transfer within each component is essential for the understanding of cross-language transfer in L2, particularly, and bilingual cognitive processing, generally. (Eighty-eight references are attached.) (RS)