22,097 Hits in 5.6 sec

2D+3D Facial Expression Recognition via Discriminative Dynamic Range Enhancement and Multi-Scale Learning [article]

Yang Jiao, Yi Niu, Trac D. Tran, Guangming Shi
2020 arXiv   pre-print
Furthermore, the constrained optimization is modeled as a K-edges maximum weight path problem in a directed acyclic graph, and we solve it efficiently via dynamic programming.  ...  Finally, we also design an efficient Facial Attention structure to automatically locate subtle discriminative facial parts for multi-scale learning, and train it with a proposed loss function ℒ_FA without  ...  We model the constrained optimization problem as finding a K-edges maximum weight path in a directed acyclic graph (DAG), and obtain the optimal solution with efficient dynamic programming.  ... 
arXiv:2011.08333v1 fatcat:wapprzdrobhlho3j3ygsckqmsq

Structural Information Learning Machinery: Learning from Observing, Associating, Optimizing, Decoding, and Abstracting [article]

Angsheng Li
2020 arXiv   pre-print
The principle and criterion of the structural information learning machines are maximization of decoding information from the data points observed together with the relationships among the data points,  ...  In the present paper, we propose the model of structural information learning machines (SiLeM for short), leading to a mathematical definition of learning by merging the theories of computation and information  ...  Structural Entropy of Graphs To develop our information theoretical model of learning, we recall the notion of structural entropy of graphs [12] .  ... 
arXiv:2001.09637v1 fatcat:nkt6efij3zcr3g7oid36gvktge

Research on Dynamic Graph Target Tracking Method Fusing the Color Local Entropy

Junchang Zhang, Chenyang Xia, Leili Hu, Yanling Zhou, Kei Eguchi, Tong Chen
2018 ITM Web of Conferences  
Firstly, to make the superpixels edge fit better and structure tighter, the local gradient feature is fused into the simple linear iterative clustering (SLIC) method.  ...  Thirdly, in order to make the proposed tracker more robust, the color local entropy is fused into the diagonal elements of the affinity matrix.  ...  However, Cai7 proposed the dynamic graph tracking (DGT) algorithm that has a good performance in dealing with deformation and occlusion problem.  ... 
doi:10.1051/itmconf/20181702004 fatcat:dmalmtj7m5axdlm5ln2szuhyzm

A Probabilistic Substructure-Based Approach for Graph Classification

H.D.K. Moonesinghe, Hamed Valiza, Samah Fodeh, Pang-Ning Tan
2007 19th IEEE International Conference on Tools with Artificial Intelligence(ICTAI 2007)  
More specifically, we use a frequent subgraph mining algorithm to construct substructure based descriptors and apply the maximum entropy principle to convert the local patterns into a global classification  ...  Empirical studies conducted on real world data sets showed that the maximum entropy substructure-based approach often outperforms existing feature vector methods using AdaBoost and Support Vector Machine  ...  George Karypis for providing the PAFI software containing FSG graph mining algorithm.  ... 
doi:10.1109/ictai.2007.159 dblp:conf/ictai/MoonesingheVFT07 fatcat:zo36nslkczafxn2wfqd5ztktsy

A Graph Data Augmentation Strategy with Entropy Preservation [article]

Xue Liu, Dan Sun, Wei Wei
2022 arXiv   pre-print
However, these operations inevitably do damage to the integrity of information structures and have to sacrifice the smoothness of feature manifold.  ...  Considering the preservation of graph entropy, we propose an effective strategy to generate randomly perturbed training data but maintain both graph topology and graph entropy.  ...  Thus graph entropy is widely used to describe and understand the dynamics of a graph quantitatively in terms of general topology or features.  ... 
arXiv:2107.06048v2 fatcat:ahudexdpbjhgnb6eekmung2zuu

Rapid Prediction of Phonon Structure and Properties using an Atomistic Line Graph Neural Network (ALIGNN) [article]

Ramya Gurunathan and Kamal Choudhary and Francesca Tavazza
2022 arXiv   pre-print
The model predictions are shown to capture the spectral features of the phonon density-of-states, effectively categorize dynamical stability, and lead to accurate predictions of DOS-derived thermal and  ...  thermodynamic properties, including heat capacity C_V, vibrational entropy S_vib, and the isotopic phonon scattering rate τ^-1_i.  ...  information about bond angle distributions has provided a means to heuristically classify local structures [15] .  ... 
arXiv:2207.12510v1 fatcat:7rudvu5hr5a6vcos4bikfb6i7q

Inferring Coupling of Distributed Dynamical Systems via Transfer Entropy [article]

Oliver M. Cliff and Mikhail Prokopenko and Robert Fitch
2016 arXiv   pre-print
In this work, we are interested in structure learning for a set of spatially distributed dynamical systems, where individual subsystems are coupled via latent variables and observed through a filter.  ...  Standard approaches to structure learning are not applicable in this framework due to the hidden variables, however we can exploit the properties of certain dynamical systems to formulate exact methods  ...  Special thanks to Joseph Lizier, Jürgen Jost, and Wolfram Martens for their incite in regards to dynamical systems.  ... 
arXiv:1611.00549v1 fatcat:dtbk3gbiozb6neb3i5uvbp3maq

Soft Hierarchical Graph Recurrent Networks for Many-Agent Partially Observable Environments [article]

Zhenhui Ye, Xiaohong Jiang, Guanghua Song, Bowei Yang
2021 arXiv   pre-print
To encourage exploration and improve robustness, we design a maximum-entropy learning method to learn stochastic policies of a configurable target action entropy.  ...  remembering its own experience, we propose a novel network structure called hierarchical graph recurrent network(HGRN) for multi-agent cooperation under partial observability.  ...  (b) We propose two maximum-entropy MADRL models that introduce a learnable temperature parameter to learn our HGRN-structured policy with a configurable target action entropy, namely Soft-HGRN and SAC-HGRN  ... 
arXiv:2109.02032v1 fatcat:aks4pnleavdglpcxmny3z2o5um

Bayesian Networks and the Imprecise Dirichlet Model Applied to Recognition Problems [chapter]

Cassio P. de Campos, Qiang Ji
2011 Lecture Notes in Computer Science  
This paper describes an Imprecise Dirichlet Model and the maximum entropy criterion to learn Bayesian network parameters under insu cient and incomplete data.  ...  It is also described how the same idea can be used to learn dynamic Bayesian networks.  ...  We focus on parameter learning in a BN where the structure (i.e. the graph) is known.  ... 
doi:10.1007/978-3-642-22152-1_14 fatcat:z74bw4kyzfaafhrq4mzwc7wviy

Improved appearance-based matching in similar and dynamic environments using a Vocabulary tree

Deon Sabatta, Davide Scaramuzza, Roland Siegwart
2010 2010 IEEE International Conference on Robotics and Automation  
Two methods of adjusting these feature entropies are proposed, one decreasing entropy related to incorrect features in a uniform manner and the other proportional to the contribution of the said feature  ...  In this paper we present a topological map building algorithm based on a Vocabulary Tree that is robust to features present in dynamic or similar environments.  ...  Through doing this we hope to be able to learn which features are typically found on dynamic objects, and exclude them from the mapping process, regardless of whether or not the object is moving when we  ... 
doi:10.1109/robot.2010.5509382 dblp:conf/icra/SabattaSS10 fatcat:ebz4saqq7zfk7fhiw3z3av5b3y

Quantifying Brain Dynamics and Structure Across Scales [article]

Ben Fulcher
system coupling Stick angle distribution Visibility graph Extreme events Local motifs Singular spectrum analysis Domain-specific techniques (Phys) Nonlinear 2D embedding structure Nonlinear  ...  Autocorrelation robustness Scaling and fluctuation analysis Permutation robustness Local extrema Seasonality tests Zero crossing rates Information Theory Information dynamics Sample Entropy  ... 
doi:10.6084/m9.figshare.15001185.v1 fatcat:kjzkzrkxjzgc5fapgtldol5rdm

Joint Learning of Graph Representation and Node Features in Graph Convolutional Neural Networks [article]

Jiaxiang Tang, Wei Hu, Xiang Gao, Zongming Guo
2019 arXiv   pre-print
In order to adapt to the underlying structure of node features in different layers, we propose dynamic learning of graphs and node features jointly in GCNNs.  ...  In particular, we cast the graph optimization problem as distance metric learning to capture pairwise similarities of features in each layer.  ...  Proposed Network Structure Having discussed the problem formulation of joint learning, we elaborate on the proposed network architecture to realize dynamic learning of graphs and node features, with focus  ... 
arXiv:1909.04931v1 fatcat:uni3l77wwred7oeohyvhfandzi

TemporalGAT: Attention-Based Dynamic Graph Representation Learning [chapter]

Ahmed Fathy, Kan Li
2020 Lecture Notes in Computer Science  
Most existing dynamic graph representation learning methods focus on modeling dynamic graphs with fixed nodes due to the complexity of modeling dynamic graphs, and therefore, cannot efficiently learn the  ...  We propose a deep attention model to learn low-dimensional feature representations which preserves the graph structure and features among series of graph snapshots over time.  ...  [25] designed two loss functions to capture the local and global graph structure.  ... 
doi:10.1007/978-3-030-47426-3_32 fatcat:njgqbwuujnapxkikqrjpimbp4q

Accelerating the identification of informative reduced representations of proteins with deep learning for graphs [article]

Federico Errica, Marco Giulini, Davide Bacciu, Roberto Menichetti, Alessio Micheli, Raffaello Potestio
2020 arXiv   pre-print
We show that deep graph networks are accurate and remarkably efficient, with a speedup factor as large as 10^5 with respect to the algorithmic computation of the mapping entropy.  ...  Here, we describe the implementation of a deep learning approach aimed at accelerating the calculation of the mapping entropy.  ...  Machine Learning Model As mentioned in Section I, the great potential of DGNs lies in the ability of efficiently handling graphs of arbitrary structure and size.  ... 
arXiv:2007.08658v1 fatcat:3a6faqwb2vf3lb3mjcigvg3abe

Advances in neuro information processing systems 11: Proceedings of the 1998 conference

1999 Computers and Mathematics with Applications  
Learning mixture hierarchies (Nuno Vasconcelos and Andrew Lippman). Discovering hidden features with Gaussian processes regression (Francesco Vivarelli and Christopher K.I. Williams).  ...  Graph matching for shape retrieval (Benoit Huet Andrew D.J. Cross and Edwin R. Han- Kernel PCA and de-noising in feature spaces (Sebastian Mika, Bernhard Sch61kopf, Alex J.  ... 
doi:10.1016/s0898-1221(99)91215-4 fatcat:rs55c6okvzgqlggua3tghun6me
« Previous Showing results 1 — 15 out of 22,097 results