D. O. Lazarev
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
D. O. Lazarev has a total of 7 co-authors. Here are the top co-authors by number of shared publications:
N.N. Kuzyrin
2 shared publications- Alexander Vladimirovich Shokurov
Ivannikov Institute for System Programming, Russian Academy of Sciences, Moscow, Russia
1 shared publication Dolgoprudnyj Mipt
1 shared publicationMoscow Isp Ras
1 shared publicationN.N. Kuzjurin
1 shared publicationN.N. Kuzyurin
Московский физико-технический институт
1 shared publication
Latest Publications
- 22 August 2026
- Published by Pleiades Publishing Ltd in Programming and Computer Software
- Vol. 52 (4) , 344-357
Data Augmentation for Machine Learning on Missing Data
The impact of missing data augmentation and class balancing methods on the prediction accuracy is studied. Results for missing and complete datasets are compared. A new algorithm called “drop-and-replace” for missing data imputation, augmentation, and class balancing is proposed. The quality of the drop-and-replace algorithm is compared to the quality of efficient and widely studied augmentation methods, such as SMOTE and mixup. For comparison, 7 datasets with sizes from 38 up to 70 000 are used. The target metric is balanced accuracy, and the nested cross-validation algorithm is used for validation. For the imputation of small datasets, the mean imputation method is used. The imputation of average or large datasets are performed using the MICE method of multivariate imputation by chained equations. The problem of training machine learning algorithms on missing data is studied in the probably approximately correct learning setting, or PAC learning. The increase of Vapnik–Chervonenkis dimension, or VC dimension, of the function set due to data imputation with one value from a finite set is evaluated. It is shown that the needed dataset size for PAC learning with sufficient accuracy, grows logarithmically slow with an increase in the filling set size. It is proved that drop-and-replace augmentation allows one to maximize balanced accuracy when learning on small linearly separable datasets containing missing values. For larger datasets, SMOTE and mixup augmentation methods are most efficient. For 8 datasets out of 11, data augmentation, which increments data size, increases classification accuracy; in 2 of 3 other cases, dataset balancing alone can improve classification accuracy; in the final case, the initial dataset is balanced and augmentation cannot increase classification accuracy. When learning on average and large datasets, in all 3 cases, augmentation more significantly improves accuracy for incomplete datasets than for complete ones. Thus, augmentation is more efficient when learning on missing data.
- 22 August 2026
- Published by Pleiades Publishing Ltd in Programming and Computer Software
- Vol. 52 (4) , 344-357
Data Augmentation for Machine Learning on Missing Data
The impact of missing data augmentation and class balancing methods on the prediction accuracy is studied. Results for missing and complete datasets are compared. A new algorithm called “drop-and-replace” for missing data imputation, augmentation, and class balancing is proposed. The quality of the drop-and-replace algorithm is compared to the quality of efficient and widely studied augmentation methods, such as SMOTE and mixup. For comparison, 7 datasets with sizes from 38 up to 70 000 are used. The target metric is balanced accuracy, and the nested cross-validation algorithm is used for validation. For the imputation of small datasets, the mean imputation method is used. The imputation of average or large datasets are performed using the MICE method of multivariate imputation by chained equations. The problem of training machine learning algorithms on missing data is studied in the probably approximately correct learning setting, or PAC learning. The increase of Vapnik–Chervonenkis dimension, or VC dimension, of the function set due to data imputation with one value from a finite set is evaluated. It is shown that the needed dataset size for PAC learning with sufficient accuracy, grows logarithmically slow with an increase in the filling set size. It is proved that drop-and-replace augmentation allows one to maximize balanced accuracy when learning on small linearly separable datasets containing missing values. For larger datasets, SMOTE and mixup augmentation methods are most efficient. For 8 datasets out of 11, data augmentation, which increments data size, increases classification accuracy; in 2 of 3 other cases, dataset balancing alone can improve classification accuracy; in the final case, the initial dataset is balanced and augmentation cannot increase classification accuracy. When learning on average and large datasets, in all 3 cases, augmentation more significantly improves accuracy for incomplete datasets than for complete ones. Thus, augmentation is more efficient when learning on missing data.
- Open Access
- 1 January 2018
- Published by Institute for System Programming of the Russian Academy of Sciences in Proceedings of the Institute for System Programming of the RAS
- Vol. 30 (6) , 105-122
On-line algorithm for scheduling parallel tasks on related computational clusters with processors of different capacities and it’s average-case analysis
Archives of Proceedings of ISP RAS Archives of Proceedings of ISP RAS
- 1 January 2018
- Published by Institute for System Programming of the Russian Academy of Sciences in Proceedings of the Institute for System Programming of the RAS
- Vol. 30 (6) , 105-122
Open AccessOn-line algorithm for scheduling parallel tasks on related computational clusters with processors of different capacities and it’s average-case analysis
Archives of Proceedings of ISP RAS Archives of Proceedings of ISP RAS
- Open Access
- 1 January 2018
- Published by Institute for System Programming of the Russian Academy of Sciences in Proceedings of the Institute for System Programming of the RAS
- Vol. 30 (4) , 209-230
On on-line algorithms for Bin, Strip and Box Packing, and their worst- and average-case analysis
Archives of Proceedings of ISP RAS Archives of Proceedings of ISP RAS
- 1 January 2018
- Published by Institute for System Programming of the Russian Academy of Sciences in Proceedings of the Institute for System Programming of the RAS
- Vol. 30 (4) , 209-230
Open AccessOn on-line algorithms for Bin, Strip and Box Packing, and their worst- and average-case analysis
Archives of Proceedings of ISP RAS Archives of Proceedings of ISP RAS
- Open Access
- 1 January 2017
- Published by Institute for System Programming of the Russian Academy of Sciences in Proceedings of the Institute for System Programming of the RAS
- Vol. 29 (6) , 213-220
Analysis of size of the largest dense subgraph of random hypergraph
- 1 January 2017
- Published by Institute for System Programming of the Russian Academy of Sciences in Proceedings of the Institute for System Programming of the RAS
- Vol. 29 (6) , 213-220
Open AccessAnalysis of size of the largest dense subgraph of random hypergraph
- Open Access
- 1 January 2017
- Published by Institute for System Programming of the Russian Academy of Sciences in Proceedings of the Institute for System Programming of the RAS
- Vol. 29 (6) , 221-228
An algorithm for Multiple Strip Package and its average case evaluation
- 1 January 2017
- Published by Institute for System Programming of the Russian Academy of Sciences in Proceedings of the Institute for System Programming of the RAS
- Vol. 29 (6) , 221-228
Open AccessAn algorithm for Multiple Strip Package and its average case evaluation