Search

Dividing ads ranking system into retrieval, early, and final stages is a common practice in large scale ads recommendation to balance the efficiency and accuracy. The early stage ranking often uses efficient models to generate candidates out of a set of retrieved ads. The candidates are then fed into a more computationally intensive but accurate final stage ranking system to produce the final ads recommendation. As the early and final stage ranking use different features and model architectures because of system constraints, a serious ranking consistency issue arises where the early stage has a low ads recall, i.e., top ads in the final stage are ranked low in the early stage. In order to pass better ads from the early to the final stage ranking, we propose a multi-task learning framework for early stage ranking to capture multiple final stage ranking components (i.e. ads clicks and ads quality events) and their task relations. With our multi-task learning framework, we can not only achieve serving cost saving from the model consolidation, but also improve the ads recall and ranking consistency. In the online A/B testing, our framework achieves significantly higher click-through rate (CTR), conversion rate (CVR), total value and better ads-quality (e.g. reduced ads cross-out rate) in a large scale industrial ads ranking system., Comment: Accepted by AdKDD 23

Disentanglement is hypothesized to be beneficial towards a number of downstream tasks. However, a common assumption in learning disentangled representations is that the data generative factors are statistically independent. As current methods are almost solely evaluated on toy datasets where this ideal assumption holds, we investigate their performance in hierarchical settings, a relevant feature of real-world data. In this work, we introduce Boxhead, a dataset with hierarchically structured ground-truth generative factors. We use this novel dataset to evaluate the performance of state-of-the-art autoencoder-based disentanglement models and observe that hierarchical models generally outperform single-layer VAEs in terms of disentanglement of hierarchically arranged factors., Comment: NeurIPS 2021 Workshop on Shared Visual Representations in Human and Machine Intelligence (SVRHM 2021)

Background: Coronavirus disease 2019 (COVID-19) and Influenza A are common disease caused by viral infection. The clinical symptoms and transmission routes of the two diseases are similar. However, there are no relevant studies on laboratory diagnostic models to discriminate COVID-19 and influenza A. This study aims at establishing a signature of laboratory findings to tell patients with COVID-19 apart from those with influenza A perfectly. Materials: In this study, 56 COVID-19 patients and 54 influenza A patients were included. Laboratory findings, epidemiological characteristics and demographic data were obtained from electronic medical record databases. Elastic network models, followed by a stepwise logistic regression model were implemented to identify indicators capable of discriminating COVID-19 and influenza A. A nomogram is diagramed to show the resulting discriminative model. Results: The majority of hematological and biochemical parameters in COVID-19 patients were significantly different from those in influenza A patients. In the final model, albumin/globulin (A/G), total bilirubin (TBIL) and erythrocyte specific volume (HCT) were selected as predictors. Using an external dataset, the model was validated to perform well. Conclusion: A diagnostic model of laboratory findings was established, in which A/G, TBIL and HCT were included as highly relevant indicators for the segmentation of COVID-19 and influenza A, providing a complimentary means for the precise diagnosis of these two diseases., Comment: 26 pages,3 figures

Body orientation estimation provides crucial visual cues in many applications, including robotics and autonomous driving. It is particularly desirable when 3-D pose estimation is difficult to infer due to poor image resolution, occlusion or indistinguishable body parts. We present COCO-MEBOW (Monocular Estimation of Body Orientation in the Wild), a new large-scale dataset for orientation estimation from a single in-the-wild image. The body-orientation labels for around 130K human bodies within 55K images from the COCO dataset have been collected using an efficient and high-precision annotation pipeline. We also validated the benefits of the dataset. First, we show that our dataset can substantially improve the performance and the robustness of a human body orientation estimation model, the development of which was previously limited by the scale and diversity of the available training data. Additionally, we present a novel triple-source solution for 3-D human pose estimation, where 3-D pose labels, 2-D pose labels, and our body-orientation labels are all used in joint training. Our model significantly outperforms state-of-the-art dual-source solutions for monocular 3-D human pose estimation, where training only uses 3-D pose labels and 2-D pose labels. This substantiates an important advantage of MEBOW for 3-D human pose estimation, which is particularly appealing because the per-instance labeling cost for body orientations is far less than that for 3-D poses. The work demonstrates high potential of MEBOW in addressing real-world challenges involving understanding human behaviors. Further information of this work is available at https://chenyanwu.github.io/MEBOW/., Comment: Accepted by CVPR2020