9,403 Hits in 5.8 sec

Towards Explanation for Unsupervised Graph-Level Representation Learning [article]

Qinghua Zheng, Jihong Wang, Minnan Luo, Yaoliang Yu, Jundong Li, Lina Yao, Xiaojun Chang
2022 arXiv   pre-print
Existing explanation methods focus on the supervised settings, , node classification and graph classification, while the explanation for unsupervised graph-level representation learning is still unexplored  ...  In this paper, we advance the Information Bottleneck principle (IB) to tackle the proposed explanation problem for unsupervised graph representations, which leads to a novel principle, Unsupervised Subgraph  ...  Unsupervised Subgraph Information Bottleneck In this paper, we study the unexplored explanation problem for unsupervised graph-level representation learning.  ... 
arXiv:2205.09934v2 fatcat:to7bd55djbdelapvygslqfxz6i

Context Matters: Graph-based Self-supervised Representation Learning for Medical Images [article]

Li Sun, Ke Yu, Kayhan Batmanghelich
2020 arXiv   pre-print
We introduce a novel approach with two levels of self-supervised representation learning objectives: one on the regional anatomical level and another on the patient-level.  ...  In addition, the graph representation has the advantage of handling any arbitrarily sized image in full resolution.  ...  Importantly, the unsupervised approach has the capability of learning robust representation since the features are not optimized towards solving a single supervised task.  ... 
arXiv:2012.06457v1 fatcat:4d5byeo2k5cm3m7w5z2knqxwzu

Effective Representation to Capture Collaboration Behaviors between Explainer and User [article]

Arjun Akula, Song-Chun Zhu
2022 arXiv   pre-print
Recently, there has been a lot of focus on building XAI models, especially to provide explanations for understanding and interpreting the predictions made by deep learning models.  ...  An explainable AI (XAI) model aims to provide transparency (in the form of justification, explanation, etc) for its predictions or actions made by it.  ...  We are trying to learn the mapping from videos to its corresponding discourse representation (i.e. discourse tree) using end-to-end learning models.  ... 
arXiv:2201.03147v1 fatcat:rnmfa6zr75da5adqm2t6lcfhiu

Foundations of Unsupervised Learning (Dagstuhl Seminar 16382)

Maria-Florina Balcan, Shai Ben-David, Ruth Urner, Ulrike Von Luxburg, Marc Herbstritt
2017 Dagstuhl Reports  
This report documents the program and the outcomes of Dagstuhl Seminar 16382 "Foundations of Unsupervised Learning". Unsupervised learning techniques are frequently used in practice of data analysis.  ...  However, there is currently little formal guidance as to how, when and to what effect to use which unsupervised learning method.  ...  One promising direction for progress towards better alignment of algorithmic objectives with application needs is the development of paradigms for interactive algorithms for such unsupervised learning  ... 
doi:10.4230/dagrep.6.9.94 dblp:journals/dagstuhl-reports/BalcanBUL16 fatcat:gliqlrxzyrbzffssk5t3udw54q

A Causal-based Framework for Multimodal Multivariate Time Series Validation Enhanced by Unsupervised Deep Learning as an Enabler for Industry 4.0 [article]

Cedric Schockaert
2020 arXiv   pre-print
A research roadmap is identified to combine causal discovery and representation learning as an enabler for unsupervised Root Cause Analysis applied to the process industry.  ...  An advanced conceptual validation framework for multimodal multivariate time series defines a multi-level contextual anomaly detection ranging from an univariate context definition, to a multimodal abstract  ...  An impressive move towards unsupervised learning is currently happening in the research community accepting the limitation of existing labels for supervised learning, due to their bias without a proper  ... 
arXiv:2008.02171v1 fatcat:7aihcefnsbamffsjqfpecy4pra

A Proposal for Common Dataset in Neural-Symbolic Reasoning Studies

Özgür Yilmaz, Artur S. d'Avila Garcez, Daniel L. Silver
2016 International Workshop on Neural-Symbolic Learning and Reasoning  
We promote and analyze the needs of a common publicly available benchmark dataset to be used for neural-symbolic studies of learning and reasoning.  ...  We would like to thank the reviewers for detailed and very beneficial comments on the paper.  ...  In a neural-symbolic system, neural networks provide the machinery for efficient computation and robust learning, while logic provides high-level representations, reasoning and explanation capabilities  ... 
dblp:conf/nesy/YilmazGS16 fatcat:qf3grff5nbbjdbszeihh5ugghy

A Survey on Explainability in Machine Reading Comprehension [article]

Mokanarangan Thayaparan, Marco Valentino, André Freitas
2020 arXiv   pre-print
We present how the representation and inference challenges evolved and the steps which were taken to tackle these challenges.  ...  In addition, we identify persisting open research questions and highlight critical directions for future work.  ...  . explanation type, explanation level, format of the background knowledge, and explanation representation.  ... 
arXiv:2010.00389v1 fatcat:jzxjysnma5ee5auvplfxxfar2u

Graph Representation Learning Beyond Node and Homophily

You Li, Bei Lin, Binli Luo, Ning Gui
2022 IEEE Transactions on Knowledge and Data Engineering  
Unsupervised graph representation learning aims to distill various graph information into a downstream task-agnostic dense vector embedding.  ...  However, existing graph representation learning approaches are designed mainly under the node homophily assumption: connected nodes tend to have similar labels and optimize performance on node-centric  ...  The most important thing is to discover the interesting and challenging problem of representing graphs from the perspective of local assortativity. This will guide our next steps.  ... 
doi:10.1109/tkde.2022.3146270 fatcat:brj5c52khjckhhq4dqb3blhjrq

Anomaly Mining – Past, Present and Future [article]

Leman Akoglu
2021 arXiv   pre-print
In this article, I focus on two areas, (1) point-cloud and (2) graph-based anomaly mining.  ...  In the following, I discuss prevalent trends on graph anomaly detection prob-lems, as well as graph neural network based techniques toward automated representation learning for complex graphs.  ...  representation learning that is better suitable for OD.  ... 
arXiv:2105.10077v2 fatcat:znvvz6ewpbdpjhnhp35kudvzlu

Learning Unsupervised Hierarchies of Audio Concepts [article]

Darius Afchar, Romain Hennequin, Vincent Guigue
2022 arXiv   pre-print
In computer vision, concept learning was therein proposed to adjust explanations to the right abstraction level (e.g. detect clinical concepts from radiographs).  ...  These methods have yet to be used for MIR. In this paper, we adapt concept learning to the realm of music, with its particularities.  ...  Unsupervised Hierarchy Mining Using knowledge graphs and taxonomies is frequent leverage in MIR [25] . Unsupervised learning of hierarchy is, however, less so.  ... 
arXiv:2207.11231v1 fatcat:mpeqzhg7cbhopngla5rf665kri

Unsupervised Keyphrase Extraction via Interpretable Neural Networks [article]

Rishabh Joshi and Vidhisha Balachandran and Emily Saldanha and Maria Glenski and Svitlana Volkova and Yulia Tsvetkov
2022 arXiv   pre-print
Prior approaches for unsupervised keyphrase extraction resort to heuristic notions of phrase importance via embedding similarities or graph centrality, requiring extensive domain expertise to develop them  ...  We show that this novel approach not only alleviates the need for ad-hoc heuristics but also achieves state-of-the-art results in unsupervised keyphrase extraction across four diverse datasets in two domains  ...  by computing an average over all the phrase-level label distributions such that l e = mean(l i ): L e = − T t=1 y t log(l e ) and computes a joint explanation and classification loss for the model as:  ... 
arXiv:2203.07640v1 fatcat:7fwzzls4w5bn7jauzcyq32i74q

Privacy and Transparency in Graph Machine Learning: A Unified Perspective [article]

Megha Khosla
2022 arXiv   pre-print
Graph Machine Learning (GraphML), whereby classical machine learning is generalized to irregular graph domains, has enjoyed a recent renaissance, leading to a dizzying array of models and their applications  ...  With its growing applicability to sensitive domains and regulations by government agencies for trustworthy AI systems, researchers have started looking into the issues of transparency and privacy of graph  ...  A few works have also been proposed to explain dense unsupervised node representations [13, 12] .  ... 
arXiv:2207.10896v1 fatcat:2rr6c75wurfwblpw7p3k664zdi

Augmented base pairing networks encode RNA-small molecule binding preferences

Carlos Oliver, Vincent Mallet, Roman Sarrazin Gendron, Vladimir Reinharz, William L Hamilton, Nicolas Moitessier, Jérôme Waldispühl
2020 Nucleic Acids Research  
We also find that pre-training with an auxiliary graph representation learning task significantly boosts performance of ligand prediction.  ...  Our work is a first attempt at bringing the scalability and generalization abilities of machine learning methods to the problem of RNA drug discovery, as well as a step towards understanding the interactions  ...  ACKNOWLEDGEMENTS The authors would like to thank Mathieu Blanchette and Jacques Boitreaud for helpful feedback and discussions.  ... 
doi:10.1093/nar/gkaa583 pmid:32652015 fatcat:gyqz4xoi7ba2to5np5wxdxtxca

Finding Interpretable Concept Spaces in Node Embeddings using Knowledge Bases [article]

Maximilian Idahl, Megha Khosla, Avishek Anand
2019 arXiv   pre-print
In this paper we propose and study the novel problem of explaining node embeddings by finding embedded human interpretable subspaces in already trained unsupervised node representation embeddings.  ...  We use an external knowledge base that is organized as a taxonomy of human-understandable concepts over entities as a guide to identify subspaces in node embeddings learned from an entity graph derived  ...  Research Questions and our Approach We propose a general approach for post-hoc interpretability of node representation learned by an unsupervised or semi-supervised method.  ... 
arXiv:1910.05030v1 fatcat:6l5pqdvesjb35c3olxkfnvyd6e

Towards an Explainable Threat Detection Tool

Alison Smith-Renner, Rob Rua, Mike Colony
2019 International Conference on Intelligent User Interfaces  
for evaluating these explanations.  ...  To this end, we have developed an unsupervised learning anomaly detection system to identify anomalous behavior without training data.  ...  This publication was cleared for public release via 88ABW-2019-0665.  ... 
dblp:conf/iui/Smith-RennerRC19 fatcat:r2za3bytizdgzjf2pclx57mdhy
« Previous Showing results 1 — 15 out of 9,403 results