George Almpanidis
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Previous Organizations / Institutes
The organizations / institutes from the scholar's other articles in the past ten years
George Almpanidis has a total of 26 co-authors. Here are the top co-authors by number of shared publications:
- Constantine Kotropoulos
Aristotle University of Thessaloniki
9 shared publications - Chongsheng Zhang
School of Computer and Information Engineering, Henan University, Kaifeng, China
4 shared publications - Gaojuan Fan
College of Computer and Information Engineering, Henan University, Kaifeng, China
4 shared publications Chongsheng Zhang
Henan University
3 shared publications- Paolo Soda
Umeå University
3 shared publications Changchang Liu
2 shared publications
Latest Publications
- Top Cited Papers
- 26 February 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Neural Networks and Learning Systems
- Vol. 36 (8) , 13670-13690
A Systematic Review on Long-Tailed Learning
Long-tailed data are a special type of multiclass imbalanced data with a very large amount of minority/tail classes that have a very significant combined influence. Long-tailed learning (LTL) aims to build high-performance models on datasets with long-tailed distributions that can identify all the classes with high accuracy, in particular the minority/tail classes. It is a cutting-edge research direction that has attracted a remarkable amount of research effort in the past few years. In this article, we present a comprehensive survey of the latest advances in long-tailed visual learning. We first propose a new taxonomy for LTL, which consists of eight different dimensions, including data balancing, neural architecture, feature enrichment, logits adjustment, loss function, bells and whistles, network optimization, and posthoc processing techniques. Based on our proposed taxonomy, we present a systematic review of LTL methods, discussing their commonalities and alignable differences. We also analyze the differences between imbalance learning and LTL. Finally, we discuss prospects and future directions in this field.
- 26 February 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Neural Networks and Learning Systems
- Vol. 36 (8) , 13670-13690
Top Cited PapersA Systematic Review on Long-Tailed Learning
Long-tailed data are a special type of multiclass imbalanced data with a very large amount of minority/tail classes that have a very significant combined influence. Long-tailed learning (LTL) aims to build high-performance models on datasets with long-tailed distributions that can identify all the classes with high accuracy, in particular the minority/tail classes. It is a cutting-edge research direction that has attracted a remarkable amount of research effort in the past few years. In this article, we present a comprehensive survey of the latest advances in long-tailed visual learning. We first propose a new taxonomy for LTL, which consists of eight different dimensions, including data balancing, neural architecture, feature enrichment, logits adjustment, loss function, bells and whistles, network optimization, and posthoc processing techniques. Based on our proposed taxonomy, we present a systematic review of LTL methods, discussing their commonalities and alignable differences. We also analyze the differences between imbalance learning and LTL. Finally, we discuss prospects and future directions in this field.
- 30 July 2022
- Published by Springer Nature in Applied Intelligence
- Vol. 53 (7) , 8506
Correction to: An empirical study on the joint impact of feature selection and data resampling on imbalance classification
- 30 July 2022
- Published by Springer Nature in Applied Intelligence
- Vol. 53 (7) , 8506
Correction to: An empirical study on the joint impact of feature selection and data resampling on imbalance classification
- 23 June 2022
- Published by Springer Nature in Applied Intelligence
- Vol. 53 (5) , 1-13
An empirical study on the joint impact of feature selection and data resampling on imbalance classification
- 23 June 2022
- Published by Springer Nature in Applied Intelligence
- Vol. 53 (5) , 1-13
An empirical study on the joint impact of feature selection and data resampling on imbalance classification
- Open Access
- 26 August 2019
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Access
- Vol. 7, 120997-121014
Gridvoronoi: An Efficient Spatial Index for Nearest Neighbor Query Processing
In this paper, based upon Voronoi Diagram, we propose GridVoronoi which is a novel spatial index that enables users to find the spatial nearest neighbour (NN) from two-dimensional (2D) datasets in almost O(1) time. GridVoronoi augments the Voronoi Diagram with a virtual grid to promptly find out (in a geometric space) which Voronoi cell contains the query point. It consists of an off-line data pre-processing phase and an on-line query processing phase. In the off-line phase, the digital geographical space is partitioned with a Voronoi Diagram and a virtual grid, respectively. Next, for each square unit (i.e., grid cell), the corresponding Voronoi cells that contain or intersect with this square are derived and kept in a hashmap-like structure. In the on-line phase, for each real-time spatial NN query, the algorithm first identifies which virtual square(s) contain(s) this query; then looks up the hashmap structure to find the corresponding Voronoi cell(s) for this grid cell and the final result for the query. Overall, GridVoronoi significantly reduces the time complexity in finding spatial NN in 2D space, thus improves the efficiency of real-time spatial NN queries and Location Based Services.
- 26 August 2019
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Access
- Vol. 7, 120997-121014
Open AccessGridvoronoi: An Efficient Spatial Index for Nearest Neighbor Query Processing
In this paper, based upon Voronoi Diagram, we propose GridVoronoi which is a novel spatial index that enables users to find the spatial nearest neighbour (NN) from two-dimensional (2D) datasets in almost O(1) time. GridVoronoi augments the Voronoi Diagram with a virtual grid to promptly find out (in a geometric space) which Voronoi cell contains the query point. It consists of an off-line data pre-processing phase and an on-line query processing phase. In the off-line phase, the digital geographical space is partitioned with a Voronoi Diagram and a virtual grid, respectively. Next, for each square unit (i.e., grid cell), the corresponding Voronoi cells that contain or intersect with this square are derived and kept in a hashmap-like structure. In the on-line phase, for each real-time spatial NN query, the algorithm first identifies which virtual square(s) contain(s) this query; then looks up the hashmap structure to find the corresponding Voronoi cell(s) for this grid cell and the final result for the query. Overall, GridVoronoi significantly reduces the time complexity in finding spatial NN in 2D space, thus improves the efficiency of real-time spatial NN queries and Location Based Services.
- 13 February 2019
- Published by Springer Nature in Neural Processing Letters
- Vol. 50 (1) , 957-987
On Incremental Learning for Gradient Boosting Decision Trees
- 13 February 2019
- Published by Springer Nature in Neural Processing Letters
- Vol. 50 (1) , 957-987
On Incremental Learning for Gradient Boosting Decision Trees
- 1 July 2018
- Published by Elsevier in Expert Systems with Applications
- Vol. 101, 91-115
An empirical evaluation of high utility itemset mining algorithms
- 1 July 2018
- Published by Elsevier in Expert Systems with Applications
- Vol. 101, 91-115
An empirical evaluation of high utility itemset mining algorithms
- Top Cited PapersOpen Access
- 1 October 2017
- Published by Elsevier in Expert Systems with Applications
- Vol. 82, 128-150
An up-to-date comparison of state-of-the-art classification algorithms
- 1 October 2017
- Published by Elsevier in Expert Systems with Applications
- Vol. 82, 128-150
Top Cited PapersOpen AccessAn up-to-date comparison of state-of-the-art classification algorithms
- Open Access
- 1 February 2009
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Audio, Speech, and Language Processing
- Vol. 17 (2) , 287-298
Robust Detection of Phone Boundaries Using Model Selection Criteria With Few Observations
Automatic phone segmentation techniques based on model selection criteria are studied. We investigate the phone boundary detection efficiency of entropy- and Bayesian- based model selection criteria in continuous speech based on the DISTBIC hybrid segmentation algorithm. DISTBIC is a text-independent bottom-up approach that identifies sequential model changes by combining metric distances with statistical hypothesis testing. Using robust statistics and small sample corrections in the baseline DISTBIC algorithm, phone boundary detection accuracy is significantly improved, while false alarms are reduced. We also demonstrate further improvement in phonemic segmentation by taking into account how the model parameters are related in the probability density functions of the underlying hypotheses as well as in the model selection via the information complexity criterion and by employing M-estimators of the model parameters. The proposed DISTBIC variants are tested on the NTIMIT database and the achieved F 1 measure is 74.7% using a 20-ms tolerance in phonemic segmentation.
- 1 February 2009
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Audio, Speech, and Language Processing
- Vol. 17 (2) , 287-298
Open AccessRobust Detection of Phone Boundaries Using Model Selection Criteria With Few Observations
Automatic phone segmentation techniques based on model selection criteria are studied. We investigate the phone boundary detection efficiency of entropy- and Bayesian- based model selection criteria in continuous speech based on the DISTBIC hybrid segmentation algorithm. DISTBIC is a text-independent bottom-up approach that identifies sequential model changes by combining metric distances with statistical hypothesis testing. Using robust statistics and small sample corrections in the baseline DISTBIC algorithm, phone boundary detection accuracy is significantly improved, while false alarms are reduced. We also demonstrate further improvement in phonemic segmentation by taking into account how the model parameters are related in the probability density functions of the underlying hypotheses as well as in the model selection via the information complexity criterion and by employing M-estimators of the model parameters. The proposed DISTBIC variants are tested on the NTIMIT database and the achieved F 1 measure is 74.7% using a 20-ms tolerance in phonemic segmentation.
- Open Access
- 1 January 2008
- Published by Elsevier in Speech Communication
- Vol. 50 (1) , 38-55
Phonemic segmentation using the generalised Gamma distribution and small sample Bayesian information criterion
- 1 January 2008
- Published by Elsevier in Speech Communication
- Vol. 50 (1) , 38-55
Open AccessPhonemic segmentation using the generalised Gamma distribution and small sample Bayesian information criterion
- 1 September 2007
- Published by Elsevier in Information Systems
- Vol. 32 (6) , 886-908
Combining text and link analysis for focused crawling—An application for vertical search engines
- 1 September 2007
- Published by Elsevier in Information Systems
- Vol. 32 (6) , 886-908
Combining text and link analysis for focused crawling—An application for vertical search engines