Filters








83,609 Hits in 2.1 sec

Neural Methods for Point-wise Dependency Estimation [article]

Yao-Hung Hubert Tsai, Han Zhao, Makoto Yamada, Louis-Philippe Morency, Ruslan Salakhutdinov
2020 arXiv   pre-print
In this work, instead of estimating the expected dependency, we focus on estimating point-wise dependency (PD), which quantitatively measures how likely two outcomes co-occur.  ...  However, MI is an aggregate statistic and cannot be used to measure point-wise dependency between different events.  ...  Point-wise Dependency Neural Estimation Our paper aims to identify the association for a pair of outcomes (x, y) ∈ X × Y by studying their point-wise dependency.  ... 
arXiv:2006.05553v4 fatcat:jzt62wzmlnav5ojusmevg4d37a

Behavioral and Neurophysiological Analyses of Dynamic Learning Processes

Wendy A. Suzuki, Emery N. Brown
2005 Behavioral & Cognitive Neuroscience Reviews  
They describe a state-space model of behavioral learning that provides accurate estimates of dynamic learning processes and a point-process filter algorithm that tracks the dynamic changes in neural activity  ...  Future challenges for these statistical methodologies and their application to the study of the brain basis of associative learning are discussed.  ...  (Brasted & Wise, 2004) , and SEF (Chen & Wise, 1995a) , we used statespace models to calculate the trial number of learning and point-process filter algorithms to estimate changes in neural activity  ... 
doi:10.1177/1534582305280030 pmid:16251726 fatcat:e4ssk3s34vg27k3xu2h3lpx6le

Deep Semantic Classification for 3D LiDAR Data [article]

Ayush Dewan, Gabriel L. Oliveira, Wolfram Burgard
2017 arXiv   pre-print
To learn the distinction between movable and non-movable points in the environment, we introduce an approach based on deep neural network and for detecting the dynamic points, we estimate pointwise motion  ...  In this paper, we propose a method for pointwise semantic classification of 3D LiDAR data into three classes: non-movable, movable and dynamic.  ...  In All of these methods, the neural network is trained for estimating bounding boxes for object detection, whereas, our network is trained for estimating pointwise objectness score; the information necessary  ... 
arXiv:1706.08355v1 fatcat:qiszg6e2qvbnzlxhj6d3lwqnhq

Coordinate-wise Control Variates for Deep Policy Gradients [article]

Yuanyi Zhong, Yuan Zhou, Jian Peng
2021 arXiv   pre-print
This paper investigates variance reduction with coordinate-wise and layer-wise control variates constructed from vector-valued baselines for neural net policies.  ...  The control variates (CV) method is widely used in policy gradient estimation to reduce the variance of the gradient estimators in practice.  ...  Conclusion We propose the coordinate-wise control variates for variance reduction in deep policy gradient methods.  ... 
arXiv:2107.04987v2 fatcat:d6dkth7kbbh4ppzimkramacvlu

HandFoldingNet: A 3D Hand Pose Estimation Network Using Multiscale-Feature Guided Folding of a 2D Hand Skeleton [article]

Wencan Cheng, Jae Hyun Park, Jong Hwan Ko
2021 arXiv   pre-print
For higher estimation accuracy, folding is guided by multi-scale features, which include both global and joint-wise local features.  ...  With increasing applications of 3D hand pose estimation in various human-computer interaction applications, convolution neural networks (CNNs) based estimation models have been actively explored.  ...  The joint-wise features help the model exploit natural spatial dependencies between the joints for better estimation performance. • We propose joint-wise local-feature guided folding to capture local features  ... 
arXiv:2108.05545v1 fatcat:3qyukvvrjzculg26lvxekiaggm

Improved Spatial-Temporal Forecasting through Modelling of Spatial Residuals in Recent History [chapter]

D. Pokrajac, Z. Obradovic
2001 Proceedings of the 2001 SIAM International Conference on Data Mining  
In such applications, models estimated on available attributes often have unsatisfactory explanatory power.  ...  Typical spatial prediction methods have been developed assuming non-uniform event-driven sampling [14] , where the objective is  ...  In addition to cases where improvements were achieved for L = 1, the proposed method with L = 0 improved the accuracy when used on data with the point-wise temporal dependence.  ... 
doi:10.1137/1.9781611972719.9 dblp:conf/sdm/PokrajacO01 fatcat:hk5gbqpjzvcbphm7lyaw6qbnre

Sparse Gaussian Markov Random Field Mixtures for Anomaly Detection

Tsuyoshi Ide, Ankush Khandelwal, Jayant Kalagnanam
2016 2016 IEEE 16th International Conference on Data Mining (ICDM)  
model • Priors: Categorical-Dirichlet for {h k i } (each sample) o Variable dependency: (iteratively) solve graphical lasso [Friedman+ 08] o Mixture components: (iteratively) point-estimated for  ...  the 1 st and 2 nd steps  The 1 st step achieves sparsity over variable dependency and mixture components o Variable dependency: (iteratively) solve graphical lasso o Mixture components: point-estimated  ... 
doi:10.1109/icdm.2016.0119 dblp:conf/icdm/IdeKK16 fatcat:2cydcu5rnnaw3l5v2qrdtz7jga

Bayesian estimation and model averaging of convolutional neural networks by hypernetwork

Kenya Ukai, Takashi Matsubara, Kuniaki Uehara
2019 Nonlinear Theory and Its Applications IEICE  
In this paper, we propose a regularization method that estimates the parameters of a large convolutional neural network as probabilistic distributions using a hypernetwork, which generates the parameters  ...  The experimental results demonstrate that our method and its model averaging outperform the commonly used maximum a posteriori estimation with L2 regularization.  ...  Bayesian neural networks BNNs are neural networks whose parameters are not point estimates but posterior distributions.  ... 
doi:10.1587/nolta.10.45 fatcat:vj5562byezgf5dfpdp3vxmdxf4

Problems of Machine Learning [chapter]

Alexei Ya. Chervonenkis
2011 Lecture Notes in Computer Science  
If the dependency is sought in the form of piece-wise linear or piece-wise continuous function, then the standard least square method cannot be applied and other tools should be used, such as artificial  ...  neural networks.  ... 
doi:10.1007/978-3-642-21786-9_5 fatcat:rpvrnvpkz5ettb2ilqqkmxv3oq

Fast Depth Estimation in a Single Image Using Lightweight Efficient Neural Network

Sangwon Kim, Jaeyeal Nam, Byoungchul Ko
2019 Sensors  
Therefore, software-based methods for estimating depth from a single image using machine learning or deep learning are emerging as new alternatives.  ...  Conventional methods re-construct scenes using feature points extracted from multiple images; however, these approaches require multiple images and thus are not easily implemented in various real-time  ...  Figure 4 . 4 Several convolution methods. (a) general convolution, (b) depth-wise convolution and (c) point-wise convolution.  ... 
doi:10.3390/s19204434 fatcat:3islyf6qhjcdnobskep5glrxk4

PlaneRecNet: Multi-Task Learning with Cross-Task Consistency for Piece-Wise Plane Detection and Reconstruction from a Single RGB Image [article]

Yaxu Xie, Fangwen Shu, Jason Rambach, Alain Pagani, Didier Stricker
2021 arXiv   pre-print
network for piece-wise planar segmentation and a depth decoder to reconstruct the scene from a single RGB image.  ...  Piece-wise 3D planar reconstruction provides holistic scene understanding of man-made environments, especially for indoor scenarios.  ...  point-wise depth supervision L RMSE .  ... 
arXiv:2110.11219v1 fatcat:z6zw5jz3tnfbzfuip52np536z4

MoBiNet: A Mobile Binary Network for Image Classification [article]

Hai Phan and Dang Huynh and Yihui He and Marios Savvides and Zhiqiang Shen
2019 arXiv   pre-print
However, training a binary network from scratch with separable depth-wise and point-wise convolutions in case of MobileNet is not trivial and prone to divergence.  ...  MobileNet and Binary Neural Networks are two among the most widely used techniques to construct deep learning models for performing a variety of tasks on mobile and embedded platforms.In this paper, we  ...  To improve this capability for MoBiNet, we proposed a novel method using K-dependency to augment correlation within separate channels.  ... 
arXiv:1907.12629v2 fatcat:cetvbtftmfd65n23ugt44n7iqa

MoBiNet: A Mobile Binary Network for Image Classification

Hai Phan, Dang Huynh, Yihui He, Marios Savvides, Zhiqiang Shen
2020 2020 IEEE Winter Conference on Applications of Computer Vision (WACV)  
However, training a binary network from scratch with separable depth-wise and point-wise convolutions in case of MobileNet is not trivial and prone to divergence.  ...  MobileNet and Binary Neural Networks are two among the most widely used techniques to construct deep learning models for performing a variety of tasks on mobile and embedded platforms.  ...  To improve this capability for MoBiNet, we proposed a novel method using K-dependency to augment correlation within separate channels.  ... 
doi:10.1109/wacv45572.2020.9093444 dblp:conf/wacv/PhanHHSS20 fatcat:snax2ohz3bgj5jrelg72gk2fqy

Spatio-Temporal Dual Graph Neural Networks for Travel Time Estimation [article]

Guangyin Jin, Huan Yan, Fuxian Li, Jincai Huang, Yong Li
2021 arXiv   pre-print
To address the above problems, a novel graph-based deep learning framework for travel time estimation is proposed in this paper, namely Spatio-Temporal Dual Graph Neural Networks (STDGNN).  ...  Travel time estimation is one of the core tasks for the development of intelligent transportation systems.  ...  Traditional Methods for Travel Time Estimation There exits a large body of traditional methods on travel time estimation.  ... 
arXiv:2105.13591v2 fatcat:rltzfzmpufc7hly6ty72q62mw4

Channel-wise pruning of neural networks with tapering resource constraint [article]

Alexey Kruglov
2018 arXiv   pre-print
We propose a new method for compute-constrained structured channel-wise pruning of convolutional neural networks.  ...  The trainable parameters of our pruning method are separate from the weights of the neural network, which allows us to avoid the interference with the neural network solver (e.g. avoid the direct dependence  ...  Our method can potentially constrain any resource that depends smoothly on the numbers of remaining channels at the pruning sites, like the number of FLOPs, size of weights, size of activations, and their  ... 
arXiv:1812.07060v1 fatcat:cxqn3vpurrh3xdw4chlcxxaub4
« Previous Showing results 1 — 15 out of 83,609 results