Filters








2,211 Hits in 1.8 sec

Bandits for BMO Functions [article]

Tianyu Wang, Cynthia Rudin
2020 arXiv   pre-print
Correspondence to: Tianyu Wang <tianyu@cs.duke.edu>. Proceedings of the 37 th International Conference on Machine Learning, Vienna, Austria, PMLR 119, 2020.  ...  Wang, T., Ye, W., Geng, D., and Rudin, C. (2019). Towards practical Lipschitz stochastic bandits. arXiv preprint arXiv:1901.09277.Wanigasekara, N. and Yu, C. (2019).  ...  Additional Proof: Proof of Lemma 4 The proof is due to Lemma 1 by Wang et al. (2019) . We present the proof for completeness. Lemma 4.  ... 
arXiv:2007.08703v1 fatcat:dq5xpb6oejc4vomkluvtrgkz3q

Distributionally Robust Prescriptive Analytics with Wasserstein Distance [article]

Tianyu Wang, Ningyuan Chen, Chun Wang
2021 arXiv   pre-print
In prescriptive analytics, the decision-maker observes historical samples of (X, Y), where Y is the uncertain problem parameter and X is the concurrent covariate, without knowing the joint distribution. Given an additional covariate observation x, the goal is to choose a decision z conditional on this observation to minimize the cost 𝔼[c(z,Y)|X=x]. This paper proposes a new distributionally robust approach under Wasserstein ambiguity sets, in which the nominal distribution of Y|X=x is
more » ... d based on the Nadaraya-Watson kernel estimator concerning the historical data. We show that the nominal distribution converges to the actual conditional distribution under the Wasserstein distance. We establish the out-of-sample guarantees and the computational tractability of the framework. Through synthetic and empirical experiments about the newsvendor problem and portfolio optimization, we demonstrate the strong performance and practical value of the proposed framework.
arXiv:2106.05724v1 fatcat:3be5q5wyyzedtcfewtou7bhhki

Instance Shadow Detection [article]

Tianyu Wang, Xiaowei Hu, Qiong Wang, Pheng-Ann Heng, Chi-Wing Fu
2020 arXiv   pre-print
Wang et al. [47] and Ding et al. [8] jointly detected and removed shadows by using multiple networks or a multi-branch network. To improve the detection performance, Le et al.  ... 
arXiv:1911.07034v2 fatcat:47kdmlx5zfbyjmmjzqnwzuuxre

HPT: Hierarchy-aware Prompt Tuning for Hierarchical Text Classification [article]

Zihan Wang, Peiyi Wang, Tianyu Liu, Yunbo Cao, Zhifang Sui, Houfeng Wang
2022 arXiv   pre-print
., 2021a; Wang et al., 2022) based on the PLM all follow this fine tuning paradigm.  ... 
arXiv:2204.13413v1 fatcat:t4mvnuf6grfipay73lshqpgkxy

Auto-CASH: Autonomous Classification Algorithm Selection with Deep Q-Network [article]

Tianyu Mu, Hongzhi Wang, Chunnan Wang, Zheng Liang
2020 arXiv   pre-print
The great amount of datasets generated by various data sources have posed the challenge to machine learning algorithm selection and hyperparameter configuration. For a specific machine learning task, it usually takes domain experts plenty of time to select an appropriate algorithm and configure its hyperparameters. If the problem of algorithm selection and hyperparameter optimization can be solved automatically, the task will be executed more efficiently with performance guarantee. Such problem
more » ... is also known as the CASH problem. Early work either requires a large amount of human labor, or suffers from high time or space complexity. In our work, we present Auto-CASH, a pre-trained model based on meta-learning, to solve the CASH problem more efficiently. Auto-CASH is the first approach that utilizes Deep Q-Network to automatically select the meta-features for each dataset, thus reducing the time cost tremendously without introducing too much human labor. To demonstrate the effectiveness of our model, we conduct extensive experiments on 120 real-world classification datasets. Compared with classical and the state-of-art CASH approaches, experimental results show that Auto-CASH achieves better performance within shorter time.
arXiv:2007.03254v1 fatcat:n372z6eionel5g4l54zl3fnxje

Support Vector Guided Softmax Loss for Face Recognition [article]

Xiaobo Wang, Shuo Wang, Shifeng Zhang, Tianyu Fu, Hailin Shi, Tao Mei
2018 arXiv   pre-print
Wang et al. [32] and Ranjan et al.  ...  Wang et al. [30] design an additive margin (AM-Softmax) loss to stabilize the optimization and have achieved promising performance. Deng et al.  ... 
arXiv:1812.11317v1 fatcat:ccmhtwhbordzpgihzxcyvuxnca

Single-Cell Classification Using Graph Convolutional Networks [article]

Tianyu Wang, Jun Bai, Sheida Nabavi
2021 bioRxiv   pre-print
AbstractBackgroundAnalyzing single-cell RNA sequencing (scRNAseq) data plays an important role in understanding the intrinsic and extrinsic cellular processes in biological and biomedical research. One significant effort in this area is the identification of cell types. With the availability of a huge amount of single cell sequencing data and discovering more and more cell types, classifying cells into known cell types has become a priority nowadays. Several methods have been introduced to
more » ... ify cells utilizing gene expression data. However, incorporating biological gene interaction networks has been proved valuable in cell classification procedures.ResultsIn this study, we propose a multimodal end-to-end deep learning model, named sigGCN, for cell classification that combines a graph convolutional network (GCN) and a neural network to exploit gene interaction networks. We used standard classification metrics to evaluate the performance of the proposed method on the within-dataset classification and the cross-dataset classification. We compared the performance of the proposed method with those of the existing cell classification tools and traditional machine learning classification methods.ConclusionsResults indicate that the proposed method outperforms other commonly used methods in terms of classification accuracy and F1 scores. This study shows that the integration of prior knowledge about gene interactions with gene expressions using GCN methodologies can extract effective features improving the performance of cell classification.
doi:10.1101/2021.06.13.448259 fatcat:7f3erervpvh23lnygrgwmlzrkm

Mis-classified Vector Guided Softmax Loss for Face Recognition [article]

Xiaobo Wang, Shifeng Zhang, Shuo Wang, Tianyu Fu, Hailin Shi, Tao Mei
2019 arXiv   pre-print
Wang et al. (Wang et al. 2017a ) and Ranjan et al.  ...  Though it has been extensively studied for decades (Wang, Guo, and Li 2015; Liu, Hu, and Wang 2019; Hu et al. 2015; Wang et al. 2018e; Chen, Deng, and Shen 2018; Liu et al. 2019a; Sun, Wang, and Tang.  ... 
arXiv:1912.00833v1 fatcat:5xzaugghdrcy3pbcjtnacuyvxi

Understanding the hardness of approximate query processing with joins [article]

Tianyu Liu, Chi Wang
2020 arXiv   pre-print
Research was done when Tianyu Liu was visiting Microsoft Research. R 2 .C 12 : 212 The first n − y entries are filled with b's except for one a.  ... 
arXiv:2010.00307v1 fatcat:4d4uknfrzjdotl3pgwd2gfq67m

Asymmetry of Raman scattering by structure variation in space

Ridong Wang, Pengyu Yuan, Meng Han, Shen Xu, Tianyu Wang, Xinwei Wang
2017 Optics Express  
We report on the discovery of asymmetries of Raman scattering along one scanning direction, between two scanning directions, and by structure variation of the sample in space. Asymmetry of Raman shift along the x direction, and the asymmetry of Raman shift and linewidth between the two scanning directions (x and y) are found for a 1210 nm diameter silica particle. The observed asymmetries are confirmed by further 2D Raman scanning of the same particle. To further explore the asymmetry of Raman
more » ... cattering, glass fibers of three diameters (0.53, 1.00, and 3.20 μm) are scanned along two directions. The asymmetry of Raman shift along each direction, the asymmetry of linewidth along the y direction, and the asymmetry of Raman shift and linewidth between the two scanning directions are discovered. Additionally, 11 nm-thick MoSe 2 nanosheets on silicon are used to discover whether an asymmetry of Raman scattering exists at the edge of the nanosheets. One edge of the nanosheet is scanned in four directions and the asymmetry of Raman scattering caused by the step variation is also detected. All the observed Raman scattering asymmetries are explained soundly by the Raman signal diffraction and image shift on the CCD detector arrays of the Raman spectrometer. In practice, to use scanning Raman for surface structure study, great measure has to be taken to consider the structure-induced asymmetries to uncover the real Raman wave number variation by intrinsic material structure. We propose a signal processing method by averaging the scanning points along four directions to eliminate the interference of the edge. This method works well to significantly suppress the asymmetries of Raman properties and uncover the real Raman signal change by structure variation.
doi:10.1364/oe.25.018378 pmid:28789324 fatcat:rbj6jwnwvbbprf2pwgdajchn5q

Imaging based on metalenses

Xiujuan Zou, Gaige Zheng, Quan Yuan, Wenbo Zang, Run Chen, Tianyue Li, Lin Li, Shuming Wang, Zhenlin Wang, Shining Zhu
2020 PhotoniX  
Furthermore, Wang et al. also presented full color imaging at almost the same time, utilizing the mechanism of incorporating every meta-element with Pancharatnam-Berry phase [79] .  ...  Distinguish from any of the above methods, Wang et al. proposed a new method of splitting the phase into two parts of wavelength-dependent and wavelength-independent phase and obtained a continuous broadband  ... 
doi:10.1186/s43074-020-00007-9 fatcat:35atzjmqnbb45mpxcfltztdck4

An Intelligent Model for Solving Manpower Scheduling Problems [article]

Lingyu Zhang and Tianyu Liu and Yunhai Wang
2021 arXiv   pre-print
When studying scheduling problems with multi-skill requirements and multi-resource constraints, Zheng, Wang, and Zheng [6] establish the teacher-learning algorithm (TLBO), combine the resource list with  ...  Evolution&Genetic Algorithm Implementation In this paper, inspired by the work finished by Wang, Yalaoui, and Dugardin [13] , two improved genetic algorithms based on the adaptive multi-dimensional input  ... 
arXiv:2105.03540v1 fatcat:umtmjtkt7zgtlhtaqnjz6itazu

Multifractal analysis of geodesic flows on surfaces without focal points [article]

Kiho Park, Tianyu Wang
2021 arXiv   pre-print
We study multifractal spectra of the geodesic flows on rank 1 surfaces without focal points. We compute the entropy of the level sets for the Lyapunov exponents and estimate its Hausdorff dimension from below. In doing so, we employ and generalize results of Burns and Gelfert.
arXiv:2104.01044v1 fatcat:qhd5sckeajgcplprvauyrh54mu

Energy and Charge Transport in 2D Atomic Layer Materials: Raman-Based Characterization

Ridong Wang, Tianyu Wang, Hamidreza Zobeiri, Dachao Li, Xinwei Wang
2020 Nanomaterials  
Wang et al. designed and employed a nanosecond ET-Raman technique to explore the temperature nonequilibrium among different phonon branches [35] .  ...  Wang et al. used one CW laser and one nanosecond laser with the same wavelength to construct the steady state heating and transient state heating [33] .  ... 
doi:10.3390/nano10091807 pmid:32927789 fatcat:jy2sby73jzevvkqxrmgcnmzo5e

Learning Navigation Costs from Demonstration in Partially Observable Environments [article]

Tianyu Wang, Vikas Dhiman, Nikolay Atanasov
2020 arXiv   pre-print
This paper focuses on inverse reinforcement learning (IRL) to enable safe and efficient autonomous navigation in unknown partially observable environments. The objective is to infer a cost function that explains expert-demonstrated navigation behavior while relying only on the observations and state-control trajectory used by the expert. We develop a cost function representation composed of two parts: a probabilistic occupancy encoder, with recurrent dependence on the observation sequence, and
more » ... cost encoder, defined over the occupancy features. The representation parameters are optimized by differentiating the error between demonstrated controls and a control policy computed from the cost encoder. Such differentiation is typically computed by dynamic programming through the value function over the whole state space. We observe that this is inefficient in large partially observable environments because most states are unexplored. Instead, we rely on a closed-form subgradient of the cost-to-go obtained only over a subset of promising states via an efficient motion-planning algorithm such as A* or RRT. Our experiments show that our model exceeds the accuracy of baseline IRL algorithms in robot navigation tasks, while substantially improving the efficiency of training and test-time inference.
arXiv:2002.11637v1 fatcat:bdlklf3235axjgwisnvyasgg4u
« Previous Showing results 1 — 15 out of 2,211 results