Filters








53,266 Hits in 3.2 sec

On Geodesic Distances and Contextual Embedding Compression for Text Classification [article]

Rishi Jha, Kai Mihata
2021 arXiv   pre-print
We investigate the efficacy of projecting contextual embedding data (BERT) onto a manifold, and using nonlinear dimensionality reduction techniques to compress these embeddings.  ...  On one dataset, we find that despite a 12-fold dimensionality reduction, the compressed embeddings performed within 0.1% of the original BERT embeddings on a downstream classification task.  ...  In addition, we would like to thank our deep learning professor Joseph Redmon for inspiring this project.  ... 
arXiv:2104.11295v1 fatcat:533ujtqwrrbp7i2ffuyi66ouqm

Deep learning of knowledge graph embeddings for semantic parsing of Twitter dialogs

Larry Heck, Hongzhao Huang
2014 2014 IEEE Global Conference on Signal and Information Processing (GlobalSIP)  
A deep neural network approach known as Deep Structured Semantic Modeling (DSSM) is used to scale the approach to learn neural embeddings for all of the concepts (pages) of Wikipedia.  ...  This paper presents a novel method to learn neural knowledge graph embeddings. The embeddings are used to compute semantic relatedness in a coherence-based semantic parser.  ...  For example, each word of a 40K-word vocabulary can be represented by a 10,306-dimentional vector using letter trigrams, giving a four-fold dimensionality reduction with few collisions.  ... 
doi:10.1109/globalsip.2014.7032187 dblp:conf/globalsip/HeckH14 fatcat:hrospi7f5ncmngv4p4bzuflhe4

Visualising Argumentation Graphs with Graph Embeddings and t-SNE [article]

Lars Malmqvist, Tommy Yuan, Suresh Manandhar
2021 arXiv   pre-print
This paper applies t-SNE, a visualisation technique familiar from Deep Neural Network research to argumentation graphs by applying it to the output of graph embeddings generated using several different  ...  It shows that such a visualisation approach can work for argumentation and show interesting structural properties of argumentation graphs, opening up paths for further research in the area.  ...  However, once an embedding has been generated for complex data types such as graphs, words, or images, it becomes possible to use dimensionality reduction techniques to visualise them effectively.  ... 
arXiv:2107.00528v1 fatcat:wrq7ftvefbdofmfqg6eyzmdgrm

On Bi-gram Graph Attributes

Thomas Konstantinovsky, Matan Mizrachi
2021 Computer and Information Science  
Furthermore, we showcase the different use-cases for the bi-gram graphs and how scalable it proves to be when dealing with large datasets.  ...  We observe a vast domain of tools and algorithms that can be developed on top of the graph representation; creating such a graph proves to be computationally cheap, and much of the heavy lifting is achieved  ...  K-core Dimensionality Reduction The K-core dimensionality reduction and noise reduction approach we proposed in section 3.2 demonstrated below shows outstanding results when used in classification-based  ... 
doi:10.5539/cis.v14n3p78 fatcat:s6q4z4kfbzghdhpj3hkearcbcm

Exploring Language Similarities with Dimensionality Reduction Technique [article]

Sangarshanan Veeraraghavan
2019 arXiv   pre-print
simple 2 dimensional plots.  ...  models that can be used for other languages, so here I have explored the idea of representing several known popular languages in a lower dimension such that their similarities can be visualized using  ...  For this task I chose Uniform Manifold Approximation and projection for dimensionality reduction.  ... 
arXiv:1902.06092v1 fatcat:xodikgzuqnglrlm5wgd6smfwnq

A New Distance Geometry Method for Constructing Word and Sentence Vectors

Leo Liberti
2020 Companion Proceedings of the Web Conference 2020  
We present a new methodology for producing low-dimensional word vectors based on distance geometry and dimensional reduction techniques.  ...  We use these word vectors in order to construct sentence vectors. We evaluate their usefulness in a sentence classification task performed by a simple artificial neural network.  ...  known methodologies for graph realization and dimensional reduction to NLP, and more specifically to graph embeddings for neural networks.  ... 
doi:10.1145/3366424.3391267 dblp:conf/www/Liberti20 fatcat:j2ov6mtpnfcgdcnijh64pym7gu

A Modified Isomap Approach to Manifold Learning in Word Spotting [chapter]

Sebastian Sudholt, Gernot A. Fink
2015 Lecture Notes in Computer Science  
In this paper, we propose a new approach to reducing the dimensionality of word image descriptors which is based on a modified version of the Isomap Manifold Learning algorithm.  ...  Word spotting is an effective paradigm for indexing document images with minimal human effort.  ...  For the George Washington dataset, the modified Isomap algorithm is able to achieve the same mAP values compared to no dimension reduction at an embedding dimensionality of 16.  ... 
doi:10.1007/978-3-319-24947-6_44 fatcat:pq4koge7yjadvdmmleelfpbhrq

Diffusion maps and coarse-graining: a unified framework for dimensionality reduction, graph partitioning, and data set parameterization

S. Lafon, A.B. Lee
2006 IEEE Transactions on Pattern Analysis and Machine Intelligence  
We provide evidence that nonlinear dimensionality reduction, clustering, and data set parameterization can be solved within one and the same framework.  ...  We show that clustering in embedding spaces is equivalent to compressing operators.  ...  Coifman for his insight and guidance, M. Maggioni and B. Nadler for contributing in the development of the diffusion framework, and Y. Keller for providing comments on the manuscript.  ... 
doi:10.1109/tpami.2006.184 pmid:16929727 fatcat:zwuu2xg6rvbsfovkvnsksgvgrm

Fast and Accurate Network Embeddings via Very Sparse Random Projection [article]

Haochen Chen, Syed Fahad Sultan, Yingtao Tian, Muhao Chen, Steven Skiena
2019 arXiv   pre-print
FastRP is proposed as a scalable algorithm for network embeddings.  ...  We observe that most network embedding methods consist of two components: construct a node similarity matrix and then apply dimension reduction techniques to this matrix.  ...  Examining the K-nearest neighbors (KNN) of a word is a common way to measure the quality of word embeddings [22] .  ... 
arXiv:1908.11512v1 fatcat:j4dblin5wfabnffadp42dmt7bm

Knowledge Fusion via Embeddings from Text, Knowledge Graphs, and Images [article]

Steffen Thoma, Achim Rettinger, Fabian Both
2017 arXiv   pre-print
Different basic fusion methods are evaluated on existing embedding approaches to show the potential of joining knowledge about certain concepts across modalities in a fused concept representation.  ...  We present a baseline approach for cross-modal knowledge fusion.  ...  Apart from simple concatenation we build on a set of existing methods like SVD and PCA for dimensionality reduction.  ... 
arXiv:1704.06084v1 fatcat:kgmi5hb4cnbsjp4ctzabqm7nzi

On Bi-gram Graph Attributes [article]

Thomas Konstantinovsky, Matan Mizrachi
2021 arXiv   pre-print
Furthermore, we showcase the different use-cases for the bi-gram graphs and how scalable it proves to be when dealing with large datasets.  ...  We observe a vast domain of tools and algorithms that can be developed on top of the graph representation; creating such a graph proves to be computationally cheap, and much of the heavy lifting is achieved  ...  Graph Coloring POS and NER Interrelations K-core Dimensionality Reduction The K-core dimensionality reduction and noise reduction approach we proposed in section 3.2 demonstrated below shows outstanding  ... 
arXiv:2107.02128v1 fatcat:bevx5liuxnehxb3lfqp2dwzgta

dg2pix: Pixel-Based Visual Analysis of Dynamic Graphs

Eren Cakmak, Dominik Jackle, Tobias Schreck, Daniel Keim
2020 2020 Visualization in Data Science (VDS)  
The graph embedding reduces the dynamic graph to a low-dimensional representation (50-300 dimensions) and learns the similarity between graphs to capture the evolving topology of the high-dimensional data  ...  The x-axis presents the temporal dimension, and the y-axis displays for each time step a graph embedding as a pixel-bar.  ...  Graph Embedding In the second step, we apply dimensionality reduction methods (graph embeddings) to all generated snapshots to learn the similarity between graphs and reduce the high-dimensional data to  ... 
doi:10.1109/vds51726.2020.00008 fatcat:yca23ponvbfcjnmuzqyyl7mdra

Exploratory Analysis of Legal Case Citation Data Using Node Embedding

Shreyansh Lodha, Rupali Wagh
2019 ICIC Express Letters  
Suitability of node embedding for application of machine learning algorithm is demonstrated by clustering the node vectors for finding similar cases.  ...  In recent years, network embedding using deep learning emerges as a promising breakthrough for analyzing networks.  ...  The authors would like to thank CHRIST (Deemed to be University) for its support pertaining to this study.  ... 
doi:10.24507/icicel.13.10.883 fatcat:wnqvah4me5hx5eabqthqe3ftw4

Network representation learning systematic review: Ancestors and current development state

Amina Amara, Mohamed Ali Hadj Taieb, Mohamed Ben Aouicha
2021 Machine Learning with Applications  
As part of machine learning techniques, graph embedding approaches are originally conceived for graphs constructed from feature represented datasets, like image dataset, in which links between nodes are  ...  Thus, we introduce a brief history of representation learning and word representation learning ancestor of network embedding.  ...  Therefore, we briefly review some of the traditional graph embedding approaches as a part of dimensionality reduction techniques.  ... 
doi:10.1016/j.mlwa.2021.100130 fatcat:axhg2gxkzfds3icebro6hlman4

A Deep Graph Embedding Network Model for Face Recognition [article]

Yufei Gan, Teng Yang, Chu He
2014 arXiv   pre-print
Secondly, many cascaded dimensionality reduction layers based on graph embedding framework are applied to GENet. Finally, a linear SVM classifier is used to classify dimension-reduced features.  ...  In this paper, we propose a new deep learning network "GENet", it combines the multi-layer network architec- ture and graph embedding framework.  ...  In the framework of graph embedding, we can use a unified view for understanding and explaining many of the popular dimensionality reduction algorithms, and a new dimensionality reduction algorithm called  ... 
arXiv:1409.7313v1 fatcat:ttz5uueycjg5xgltyexfv55l7q
« Previous Showing results 1 — 15 out of 53,266 results