Kenneth Mak
Research Subjects
Kenneth Mak has a total of 16 co-authors. Here are the top co-authors by number of shared publications:
- Alexandru A. Ormenisan
Hopsworks AB, Stockholm, Sweden
1 shared publication - Antonios Kouzoupis
Hopsworks AB, Stockholm, Sweden
1 shared publication - Ayushman Khazanchi
KTH Royal Institute of Technology, Stockholm, Sweden
1 shared publication - Davit Bzhalava
Department of Laboratory Medicine, Karolinska Institutet, Stockholm, Sweden
1 shared publication - Dhananjay Mukhedkar
Hopsworks AB, Stockholm, Sweden
1 shared publication - Fabio Buso
Hopsworks AB, Stockholm, Sweden
1 shared publication
Latest Publications
- Open Access
- 9 June 2024
- Published by Association for Computing Machinery (ACM)
- p. 135-147
The Hopsworks Feature Store for Machine Learning
Data management is the most challenging aspect of building Machine Learning (ML) systems. ML systems can read large volumes of historical data when training models, but inference workloads are more varied, depending on whether it is a batch or online ML system. The feature store for ML has recently emerged as a single data platform for managing ML data throughout the ML lifecycle, from feature engineering to model training to inference. In this paper, we present the Hopsworks feature store for machine learning as a highly available platform for managing feature data with API support for columnar, row-oriented, and similarity search query workloads. We introduce and address challenges solved by the feature stores related to feature reuse, how to organize data transformations, and how to ensure correct and consistent data between feature engineering, model training, and model inference. We present the engineering challenges in building high-performance query services for a feature store and show how Hopsworks outperforms existing cloud feature stores for training and online inference query workloads.
- 9 June 2024
- Published by Association for Computing Machinery (ACM)
- p. 135-147
Open AccessThe Hopsworks Feature Store for Machine Learning
Data management is the most challenging aspect of building Machine Learning (ML) systems. ML systems can read large volumes of historical data when training models, but inference workloads are more varied, depending on whether it is a batch or online ML system. The feature store for ML has recently emerged as a single data platform for managing ML data throughout the ML lifecycle, from feature engineering to model training to inference. In this paper, we present the Hopsworks feature store for machine learning as a highly available platform for managing feature data with API support for columnar, row-oriented, and similarity search query workloads. We introduce and address challenges solved by the feature stores related to feature reuse, how to organize data transformations, and how to ensure correct and consistent data between feature engineering, model training, and model inference. We present the engineering challenges in building high-performance query services for a feature store and show how Hopsworks outperforms existing cloud feature stores for training and online inference query workloads.
- 1 December 2005
- Published by Taylor & Francis in International Journal of Production Research
- Vol. 43 (23) , 4929-4944
Optimal inventory control of lumpy demand items using (s,S) policies with a maximum issue quantity restriction and opportunistic replenishments
This paper presents a modified (s, S) inventory model which describes the characteristics of an inventory system with lumpy demand items. A maximum issue quantity restriction of w units and a critical inventory position of A units are incorporated into the inventory control policy. Customer orders with demand sizes larger than the maximum issue quantity will be filtered out from the inventory system and satisfied by using special replenishment orders in order to avoid disruption to the inventory system. The option of opportunistic replenishments is introduced to further minimize the total replenishment cost. An opportunistic replenishment is initiated if the level of the current inventory position is equal to or below the critical level when a customer demand with a size exceeding the maximum issue quantity arrives, which does not only initiate a direct shipment to the customer, but also raises the inventory position to S. Two effective algorithms are developed to determine the optimal values of w, A, s and S simultaneously. The first algorithm is based on the branch-and-bound tree search technique, and the second one is based on the concept of genetic algorithms. Numerical examples are used to illustrate the effectiveness of the algorithms developed. The effects of changes in the cost and system parameters on the optimal inventory control policy are also studied by using sensitivity analysis.
- 1 December 2005
- Published by Taylor & Francis in International Journal of Production Research
- Vol. 43 (23) , 4929-4944
Optimal inventory control of lumpy demand items using (s,S) policies with a maximum issue quantity restriction and opportunistic replenishments
This paper presents a modified (s, S) inventory model which describes the characteristics of an inventory system with lumpy demand items. A maximum issue quantity restriction of w units and a critical inventory position of A units are incorporated into the inventory control policy. Customer orders with demand sizes larger than the maximum issue quantity will be filtered out from the inventory system and satisfied by using special replenishment orders in order to avoid disruption to the inventory system. The option of opportunistic replenishments is introduced to further minimize the total replenishment cost. An opportunistic replenishment is initiated if the level of the current inventory position is equal to or below the critical level when a customer demand with a size exceeding the maximum issue quantity arrives, which does not only initiate a direct shipment to the customer, but also raises the inventory position to S. Two effective algorithms are developed to determine the optimal values of w, A, s and S simultaneously. The first algorithm is based on the branch-and-bound tree search technique, and the second one is based on the concept of genetic algorithms. Numerical examples are used to illustrate the effectiveness of the algorithms developed. The effects of changes in the cost and system parameters on the optimal inventory control policy are also studied by using sensitivity analysis.