Carnegie Mellon University

About: Carnegie Mellon University is a education organization based out in Pittsburgh, Pennsylvania, United States. It is known for research contribution in the topics: Computer science & Robot. The organization has 36317 authors who have published 104359 publications receiving 5975734 citations. The organization is also known as: CMU & Carnegie Mellon.

XLNet: Generalized Autoregressive Pretraining for Language Understanding

Abstract: With the capability of modeling bidirectional contexts, denoising autoencoding based pretraining like BERT achieves better performance than pretraining approaches based on autoregressive language modeling. However, relying on corrupting the input with masks, BERT neglects dependency between the masked positions and suffers from a pretrain-finetune discrepancy. In light of these pros and cons, we propose XLNet, a generalized autoregressive pretraining method that (1) enables learning bidirectional contexts by maximizing the expected likelihood over all permutations of the factorization order and (2) overcomes the limitations of BERT thanks to its autoregressive formulation. Furthermore, XLNet integrates ideas from Transformer-XL, the state-of-the-art autoregressive model, into pretraining. Empirically, under comparable experiment setting, XLNet outperforms BERT on 20 tasks, often by a large margin, including question answering, natural language inference, sentiment analysis, and document ranking.

A Dynamic Process Model of Service Quality: From Expectations to Behavioral Intentions:

Abstract: Relying on a Bayesian-like framework, the authors develop a behavioral process model of perceived service quality. Perceptions of the dimensions of service quality are viewed to be a function of a ...

Layer-dependent ferromagnetism in a van der Waals crystal down to the monolayer limit

Abstract: Magneto-optical Kerr effect microscopy is used to show that monolayer chromium triiodide is an Ising ferromagnet with out-of-plane spin orientation. The question of what happens to the properties of a material when it is thinned down to atomic-scale thickness has for a long time been a largely hypothetical one. In the past decade, new experimental methods have made it possible to isolate and measure a range of two-dimensional structures, enabling many theoretical predictions to be tested. But it has been a particular challenge to observe intrinsic magnetic effects, which could shed light on the longstanding fundamental question of whether intrinsic long-range magnetic order can robustly exist in two dimensions. In this issue of Nature, two groups address this challenge and report ferromagnetism in atomically thin crystals. Xiang Zhang and colleagues measured atomic layers of Cr2Ge2Te6 and observed ferromagnetic ordering with a transition temperature that, unusually, can be controlled using small magnetic fields. Xiaodong Xu and colleagues measured atomic layers of CrI3 and observed ferromagnetic ordering that, remarkably, was suppressed in double layers of CrI3, but restored in triple layers. The two studies demonstrate a platform with which to test fundamental properties of purely two-dimensional magnets. Since the discovery of graphene1, the family of two-dimensional materials has grown, displaying a broad range of electronic properties. Recent additions include semiconductors with spin–valley coupling2, Ising superconductors3,4,5 that can be tuned into a quantum metal6, possible Mott insulators with tunable charge-density waves7, and topological semimetals with edge transport8,9. However, no two-dimensional crystal with intrinsic magnetism has yet been discovered10,11,12,13,14; such a crystal would be useful in many technologies from sensing to data storage15. Theoretically, magnetic order is prohibited in the two-dimensional isotropic Heisenberg model at finite temperatures by the Mermin–Wagner theorem16. Magnetic anisotropy removes this restriction, however, and enables, for instance, the occurrence of two-dimensional Ising ferromagnetism. Here we use magneto-optical Kerr effect microscopy to demonstrate that monolayer chromium triiodide (CrI3) is an Ising ferromagnet with out-of-plane spin orientation. Its Curie temperature of 45 kelvin is only slightly lower than that of the bulk crystal, 61 kelvin, which is consistent with a weak interlayer coupling. Moreover, our studies suggest a layer-dependent magnetic phase, highlighting thickness-dependent physical properties typical of van der Waals crystals17,18,19. Remarkably, bilayer CrI3 displays suppressed magnetization with a metamagnetic effect20, whereas in trilayer CrI3 the interlayer ferromagnetism observed in the bulk crystal is restored. This work creates opportunities for studying magnetism by harnessing the unusual features of atomically thin materials, such as electrical control for realizing magnetoelectronics12, and van der Waals engineering to produce interface phenomena15.

Realtime Multi-Person 2D Pose Estimation using Part Affinity Fields

Abstract: We present an approach to efficiently detect the 2D pose of multiple people in an image. The approach uses a nonparametric representation, which we refer to as Part Affinity Fields (PAFs), to learn to associate body parts with individuals in the image. The architecture encodes global context, allowing a greedy bottom-up parsing step that maintains high accuracy while achieving realtime performance, irrespective of the number of people in the image. The architecture is designed to jointly learn part locations and their association via two branches of the same sequential prediction process. Our method placed first in the inaugural COCO 2016 keypoints challenge, and significantly exceeds the previous state-of-the-art result on the MPII Multi-Person benchmark, both in performance and efficiency.