Yang Shi 石钖
Graduate student at UCIrvine
LATEST PROJECTS
![cs](https://static.wixstatic.com/media/6517ce_477fae185f4f472f9ad049559348634a~mv2.png/v1/fill/w_99,h_10,al_c,q_85,usm_0.66_1.00_0.01,blur_2,enc_auto/6517ce_477fae185f4f472f9ad049559348634a~mv2.png)
![ms-matrix](https://static.wixstatic.com/media/6517ce_db626843e1ad4ba0affd1f21082bfd0d~mv2.png/v1/fill/w_78,h_29,al_c,q_85,usm_0.66_1.00_0.01,blur_2,enc_auto/6517ce_db626843e1ad4ba0affd1f21082bfd0d~mv2.png)
![cs](https://static.wixstatic.com/media/6517ce_477fae185f4f472f9ad049559348634a~mv2.png/v1/fill/w_99,h_10,al_c,q_85,usm_0.66_1.00_0.01,blur_2,enc_auto/6517ce_477fae185f4f472f9ad049559348634a~mv2.png)
![ms-matrix](https://static.wixstatic.com/media/6517ce_db626843e1ad4ba0affd1f21082bfd0d~mv2.png/v1/fill/w_78,h_29,al_c,q_85,usm_0.66_1.00_0.01,blur_2,enc_auto/6517ce_db626843e1ad4ba0affd1f21082bfd0d~mv2.png)
Project | 01
Multi-dimensional Count Sketch
Appeared in NeurIPS 2018 workshop
When dealing with large-scale multi-dimensional data, it is crucial to design efficient compression schemes that remove redundancy while preserving relevant information. In this work, we propose multi-dimensional count sketch (MS), an unbiased estimator of the given tensor.
![qtype-concat](https://static.wixstatic.com/media/6517ce_6cb85b16a058429489c39bbf4d530009~mv2.png/v1/fill/w_120,h_92,al_c,q_85,usm_0.66_1.00_0.01,blur_2,enc_auto/6517ce_6cb85b16a058429489c39bbf4d530009~mv2.png)
![MCB-att-qtype-sim-new](https://static.wixstatic.com/media/6517ce_8ef0d2a10c0746f1bf52667fe9f96e12~mv2_d_3285_1406_s_2.png/v1/fill/w_49,h_21,al_c,q_85,usm_0.66_1.00_0.01,blur_2,enc_auto/6517ce_8ef0d2a10c0746f1bf52667fe9f96e12~mv2_d_3285_1406_s_2.png)
![qtype-concat](https://static.wixstatic.com/media/6517ce_6cb85b16a058429489c39bbf4d530009~mv2.png/v1/fill/w_120,h_92,al_c,q_85,usm_0.66_1.00_0.01,blur_2,enc_auto/6517ce_6cb85b16a058429489c39bbf4d530009~mv2.png)
![MCB-att-qtype-sim-new](https://static.wixstatic.com/media/6517ce_8ef0d2a10c0746f1bf52667fe9f96e12~mv2_d_3285_1406_s_2.png/v1/fill/w_49,h_21,al_c,q_85,usm_0.66_1.00_0.01,blur_2,enc_auto/6517ce_8ef0d2a10c0746f1bf52667fe9f96e12~mv2_d_3285_1406_s_2.png)
Project | 02
Question Type Guided Attention in Visual Question Answering
Appeared in ECCV 2018
We propose Question Type-guided Attention. It utilizes the information of question type to dynamically balance between bottom-up and top-down visual features, respectively extracted from
ResNet and Faster R-CNN networks.
[Paper] [Poster] [Codes] [Tutorial on VQA]
![Luscious Palm Leaves](https://static.wixstatic.com/media/13e37eb427f640a694d75f86537b3b55.jpg/v1/fill/w_147,h_98,al_c,q_80,usm_0.66_1.00_0.01,blur_2,enc_auto/13e37eb427f640a694d75f86537b3b55.jpg)
![Lily Pad Pond](https://static.wixstatic.com/media/ae505cb9630a240ff39bd7e3a6a30e5b.jpg/v1/fill/w_126,h_95,al_c,q_80,usm_0.66_1.00_0.01,blur_2,enc_auto/ae505cb9630a240ff39bd7e3a6a30e5b.jpg)
![Luscious Palm Leaves](https://static.wixstatic.com/media/13e37eb427f640a694d75f86537b3b55.jpg/v1/fill/w_147,h_98,al_c,q_80,usm_0.66_1.00_0.01,blur_2,enc_auto/13e37eb427f640a694d75f86537b3b55.jpg)
![Lily Pad Pond](https://static.wixstatic.com/media/ae505cb9630a240ff39bd7e3a6a30e5b.jpg/v1/fill/w_126,h_95,al_c,q_80,usm_0.66_1.00_0.01,blur_2,enc_auto/ae505cb9630a240ff39bd7e3a6a30e5b.jpg)
Project | 03
We consider the problem of robust PCA in the streaming setting with space constraints. Our result is the first to obtain finite-sample guarantees while having the weakest assumption on the sparse perturbation, namely, deterministic support, and a standard identifiability assumption on the low-rank component, namely, incoherence.
Streaming Robust PCA
Appeared in NIPS 2017 OPT Workshop
In this project, we find a intuitive and efficient way to do tensor contraction with extended BLAS kernels (Bateched BLAS). We also benchmarked with state-of-art library: BTAS.
Project | 04
![tensorcontraction](https://static.wixstatic.com/media/6517ce_0058f25ffc424eec83d16de9234572db~mv2.png/v1/fill/w_79,h_19,al_c,q_85,usm_0.66_1.00_0.01,blur_2,enc_auto/6517ce_0058f25ffc424eec83d16de9234572db~mv2.png)
![Screen Shot 2017-11-28 at 6.30.18 PM](https://static.wixstatic.com/media/6517ce_b1f4ea18ea404f89bb912ebbc827956c~mv2.png/v1/fill/w_76,h_28,al_c,q_85,usm_0.66_1.00_0.01,blur_2,enc_auto/6517ce_b1f4ea18ea404f89bb912ebbc827956c~mv2.png)
![tensorcontraction](https://static.wixstatic.com/media/6517ce_0058f25ffc424eec83d16de9234572db~mv2.png/v1/fill/w_79,h_19,al_c,q_85,usm_0.66_1.00_0.01,blur_2,enc_auto/6517ce_0058f25ffc424eec83d16de9234572db~mv2.png)
![Screen Shot 2017-11-28 at 6.30.18 PM](https://static.wixstatic.com/media/6517ce_b1f4ea18ea404f89bb912ebbc827956c~mv2.png/v1/fill/w_76,h_28,al_c,q_85,usm_0.66_1.00_0.01,blur_2,enc_auto/6517ce_b1f4ea18ea404f89bb912ebbc827956c~mv2.png)
Tensor Contraction using Extended BLAS Kernels
Appeared in HiPC 2016
We applied matrix robust Principle Component Analysis(PCA) to tensor cases: separate a tensor as a low rank tensor and a sparse tensor. This is done by using nonconvex optimization. We comapred our method with matrix robust PCA.
Project | 04
![Screen Shot 2017-11-28 at 11.42.08 PM](https://static.wixstatic.com/media/6517ce_3658b0fd03e844abb75fe423758ac652~mv2.png/v1/fill/w_49,h_7,al_c,q_85,usm_0.66_1.00_0.01,blur_2,enc_auto/6517ce_3658b0fd03e844abb75fe423758ac652~mv2.png)
![Screen Shot 2017-11-28 at 6.37.39 PM](https://static.wixstatic.com/media/6517ce_07c1936b27e54b729cb68170d32dc9c6~mv2.png/v1/fill/w_119,h_44,al_c,q_85,usm_0.66_1.00_0.01,blur_2,enc_auto/6517ce_07c1936b27e54b729cb68170d32dc9c6~mv2.png)
![Screen Shot 2017-11-28 at 11.42.08 PM](https://static.wixstatic.com/media/6517ce_3658b0fd03e844abb75fe423758ac652~mv2.png/v1/fill/w_49,h_7,al_c,q_85,usm_0.66_1.00_0.01,blur_2,enc_auto/6517ce_3658b0fd03e844abb75fe423758ac652~mv2.png)
![Screen Shot 2017-11-28 at 6.37.39 PM](https://static.wixstatic.com/media/6517ce_07c1936b27e54b729cb68170d32dc9c6~mv2.png/v1/fill/w_119,h_44,al_c,q_85,usm_0.66_1.00_0.01,blur_2,enc_auto/6517ce_07c1936b27e54b729cb68170d32dc9c6~mv2.png)
Project | 05
Tensor Robust PCA
Appeared in AISTATS 2016
Project | 06
![Plant Shed](https://static.wixstatic.com/media/fbe82a58299c4c86ab85091eeaee41ae.jpg/v1/fill/w_147,h_98,al_c,q_80,usm_0.66_1.00_0.01,blur_2,enc_auto/fbe82a58299c4c86ab85091eeaee41ae.jpg)
![Cactus Collection](https://static.wixstatic.com/media/cef4239941754da691c18a488a520181.jpg/v1/fill/w_147,h_98,al_c,q_80,usm_0.66_1.00_0.01,blur_2,enc_auto/cef4239941754da691c18a488a520181.jpg)
![Plant Shed](https://static.wixstatic.com/media/fbe82a58299c4c86ab85091eeaee41ae.jpg/v1/fill/w_147,h_98,al_c,q_80,usm_0.66_1.00_0.01,blur_2,enc_auto/fbe82a58299c4c86ab85091eeaee41ae.jpg)
![Cactus Collection](https://static.wixstatic.com/media/cef4239941754da691c18a488a520181.jpg/v1/fill/w_147,h_98,al_c,q_80,usm_0.66_1.00_0.01,blur_2,enc_auto/cef4239941754da691c18a488a520181.jpg)
Using tensor convolutional decomposition to find out usefull features for unsupervise/supervise learning.
Convolutional Tenor Decomposition in Unsupervise/Supervise Learning
Under preparation
![Plant Shed](https://static.wixstatic.com/media/fbe82a58299c4c86ab85091eeaee41ae.jpg/v1/fill/w_147,h_98,al_c,q_80,usm_0.66_1.00_0.01,blur_2,enc_auto/fbe82a58299c4c86ab85091eeaee41ae.jpg)
![Cactus Collection](https://static.wixstatic.com/media/cef4239941754da691c18a488a520181.jpg/v1/fill/w_147,h_98,al_c,q_80,usm_0.66_1.00_0.01,blur_2,enc_auto/cef4239941754da691c18a488a520181.jpg)
![Plant Shed](https://static.wixstatic.com/media/fbe82a58299c4c86ab85091eeaee41ae.jpg/v1/fill/w_147,h_98,al_c,q_80,usm_0.66_1.00_0.01,blur_2,enc_auto/fbe82a58299c4c86ab85091eeaee41ae.jpg)
![Cactus Collection](https://static.wixstatic.com/media/cef4239941754da691c18a488a520181.jpg/v1/fill/w_147,h_98,al_c,q_80,usm_0.66_1.00_0.01,blur_2,enc_auto/cef4239941754da691c18a488a520181.jpg)
Considering the physical water network as a graph, a CRF model is built and learned by the Structured Support Vector Machine (SSVM) . After that, we form high order CRF system by enforcing twitter leakage detection information.
Project | 07
Leak Event Identification in Water Systems Using High Order CRF
Course project