Tongji University

About: Tongji University is a education organization based out in Shanghai, China. It is known for research contribution in the topics: Computer science & Population. The organization has 76116 authors who have published 81176 publications receiving 1248911 citations. The organization is also known as: Tongji & Tóngjì Dàxué.

Simplified method for consolidation rate of stone column reinforced foundations

Abstract: Field observations and numerical studies demonstrated that stone columns could accelerate the rate of consolidation of soft clays. A simplified method for computing the rate of consolidation is presented in this paper by assuming that stone columns; (1) are free draining; (2) have higher drained elastic modulus than soft clay; and (3) are deformed 1D. The formats of the final solutions in vertical and radial flows are similar to those of the Terzaghi 1D solution and the Barron solution for drain wells in fine-grained soils, respectively. Modified coefficients of consolidation are introduced to account for effects of the stone column-soil modular ratio. The new solutions demonstrate stress transfer from the soil to stone columns and dissipation of excess pore water pressures due to drainage and vertical stress reduction during the consolidation. Comparisons between the results from this simplified method and the numerical study by Balaam and Booker in 1981 exhibit reasonable agreement, when the stress concentration ratio is in the practical range (2-6). The discrepancies in the results from these two methods are discussed. This paper also includes design charts and a design example.

Landscape of tumor-infiltrating T cell repertoire of human cancers

Abstract: We developed a computational method to infer the complementarity-determining region 3 (CDR3) sequences of tumor-infiltrating T cells in 9,142 RNA-seq samples across 29 cancer types. We identified over 600,000 CDR3 sequences, including 15% that were full length. CDR3 sequence length distribution and amino acid conservation, as well as variable gene usage, for infiltrating T cells in many tumors, except in brain and kidney cancers, resembled those for peripheral blood cells from healthy donors. We observed a strong association between T cell diversity and tumor mutation load, and we predicted SPAG5 and TSSK6 as putative immunogenic cancer/testis antigens in multiple cancers. Finally, we identified three potential immunogenic somatic mutations on the basis of their co-occurrence with CDR3 sequences. One of them, a PRAMEF4 mutation encoding p.Phe300Val, was predicted to result in peptide binding strongly to both MHC class I and class II molecules, with matched HLA types in its carriers. Our analyses have the potential to simultaneously identify immunogenic neoantigens and tumor-reactive T cell clonotypes.

SR-SIM: A fast and high performance IQA index based on spectral residual

Abstract: Automatic image quality assessment (IQA) attempts to use computational models to measure the image quality in consistency with subjective ratings. In the past decades, dozens of IQA models have been proposed. Though some of them can predict subjective image quality accurately, their computational costs are usually very high. To meet real-time requirements, in this paper, we propose a novel fast and effective IQA index, namely spectral residual based similarity (SR-SIM), based on a specific visual saliency model, spectral residual visual saliency. SR-SIM is designed based on the hypothesis that an image's visual saliency map is closely related to its perceived quality. Extensive experiments conducted on three large-scale IQA datasets indicate that SR-SIM could achieve better prediction performance than the other state-of-the-art IQA indices evaluated. Moreover, SR-SIM can have a quite low computational complexity. The Matlab source code of SR-SIM and the evaluation results are available online at http://sse.tongji.edu.cn/linzhang/IQA/SR-SIM/SR-SIM.htm.

SOLOv2: Dynamic and Fast Instance Segmentation

Abstract: In this work, we aim at building a simple, direct, and fast instance segmentation framework with strong performance. We follow the principle of the SOLO method of Wang et al. "SOLO: segmenting objects by locations". Importantly, we take one step further by dynamically learning the mask head of the object segmenter such that the mask head is conditioned on the location. Specifically, the mask branch is decoupled into a mask kernel branch and mask feature branch, which are responsible for learning the convolution kernel and the convolved features respectively. Moreover, we propose Matrix NMS (non maximum suppression) to significantly reduce the inference time overhead due to NMS of masks. Our Matrix NMS performs NMS with parallel matrix operations in one shot, and yields better results. We demonstrate a simple direct instance segmentation system, outperforming a few state-of-the-art methods in both speed and accuracy. A light-weight version of SOLOv2 executes at 31.3 FPS and yields 37.1% AP. Moreover, our state-of-the-art results in object detection (from our mask byproduct) and panoptic segmentation show the potential to serve as a new strong baseline for many instance-level recognition tasks besides instance segmentation. Code is available at: this https URL