Yaron Kanza
Research Subjects
Yaron Kanza has a total of 23 co-authors. Here are the top co-authors by number of shared publications:
- Divesh Srivastava
AT&T Chief Data Office, Bedminster, USA
4 shared publications - Velin Kounev
AT&T Labs-Research, United States
3 shared publications - Krystian Czapiga
AT&T Labs-Research, United States
2 shared publications - Serkan Isci
AT&T Labs-Research, United States
2 shared publications Yehoshua Sagiv
The Hebrew University
2 shared publications- Ari Biswas
University of Warwick, Coventry, United Kingdom
1 shared publication
Latest Publications
- Open Access
- 24 January 2026
- Published by Association for Computing Machinery (ACM) in Journal of Data and Information Quality
- Vol. 18 (3) , 1-29
Variable-Length Anomalous Intervals in Data Streams
Anomaly-management tools monitor data streams in complex systems, e.g., tracking weather parameters collected from sensors in a region or inspecting resource usage of machines in a data center. Their goal is to detect outliers and abnormal behavior of the system, including a burst of outliers, which is often the result of a major event, like a hurricane or a crash of machines in a data center. An interval alert is an alert raised for such a burst of outliers. In data pipelines, such alerts indicate time intervals during which major data quality issues may occur. This allows alerting downstream applications on potential events and data quality risks. In this paper we present Ourea , a novel unsupervised learning method that computes interval alerts by identifying anomalous time intervals of variable lengths in data streams. Ourea raises interval alerts on significant events, based on a concentration of outliers detected in the raw data stream and in high-level aggregate views. It uses Kernel Density Estimate (KDE) to identify the most significant time intervals and their boundaries, can issue preliminary alerts early before intervals end, and minimizes the number of raised alerts by alerting only on significant ones. Extensive experiments over real and synthetic data demonstrate the effectiveness and scalability of Ourea and show that for detecting variable-length anomalous intervals in data streams, our algorithm is more accurate and more efficient than state-of-the-art methods.
- 24 January 2026
- Published by Association for Computing Machinery (ACM) in Journal of Data and Information Quality
- Vol. 18 (3) , 1-29
Open AccessVariable-Length Anomalous Intervals in Data Streams
Anomaly-management tools monitor data streams in complex systems, e.g., tracking weather parameters collected from sensors in a region or inspecting resource usage of machines in a data center. Their goal is to detect outliers and abnormal behavior of the system, including a burst of outliers, which is often the result of a major event, like a hurricane or a crash of machines in a data center. An interval alert is an alert raised for such a burst of outliers. In data pipelines, such alerts indicate time intervals during which major data quality issues may occur. This allows alerting downstream applications on potential events and data quality risks. In this paper we present Ourea , a novel unsupervised learning method that computes interval alerts by identifying anomalous time intervals of variable lengths in data streams. Ourea raises interval alerts on significant events, based on a concentration of outliers detected in the raw data stream and in high-level aggregate views. It uses Kernel Density Estimate (KDE) to identify the most significant time intervals and their boundaries, can issue preliminary alerts early before intervals end, and minimizes the number of raised alerts by alerting only on significant ones. Extensive experiments over real and synthetic data demonstrate the effectiveness and scalability of Ourea and show that for detecting variable-length anomalous intervals in data streams, our algorithm is more accurate and more efficient than state-of-the-art methods.
- Open Access
- 3 November 2025
- Published by Association for Computing Machinery (ACM)
- p. 19-21
Geospatially Enhanced Quantum Optimization for Large-Scale Network Planning: A Summary
- 3 November 2025
- Published by Association for Computing Machinery (ACM)
- p. 19-21
Open AccessGeospatially Enhanced Quantum Optimization for Large-Scale Network Planning: A Summary
- Open Access
- 3 November 2025
- Published by Association for Computing Machinery (ACM)
- p. 45-55
Decentralized Trajectory Matching for Revealing Unauthorized Stalking by Bluetooth Trackers
- 3 November 2025
- Published by Association for Computing Machinery (ACM)
- p. 45-55
Open AccessDecentralized Trajectory Matching for Revealing Unauthorized Stalking by Bluetooth Trackers
- Open Access
- 3 November 2025
- Published by Association for Computing Machinery (ACM)
- p. 764-768
Spatially Enhanced Quantum Optimization for Network Planning
- 3 November 2025
- Published by Association for Computing Machinery (ACM)
- p. 764-768
Open AccessSpatially Enhanced Quantum Optimization for Network Planning
- Open Access
- 4 November 2024
- Published by Association for Computing Machinery (ACM) in Proceedings of the ACM on Management of Data
- Vol. 2 (5) , 1-25
Differentially Private Hierarchical Heavy Hitters
The task of finding Hierarchical Heavy Hitters (HHH) was introduced by Cormode et al. [12] as a generalisation of the heavy hitter problem. While finding HHH in data streams has been studied extensively, the question of releasing HHH when the underlying data is private remains unexplored. In this paper, we formalise and study the notion of differentially private HHH, in both the streaming and non-streaming setting. In the non-streaming setting, we show the surprising result that the relative error in estimating the count for any prefix is independent of the height of the hierarchy and the number of heavy hitters in the stream. Additionally, our algorithms also improve the error guarantees of Ghazi et al. [24] for the problem of counting over trees. Meanwhile, in the streaming setting, the main issue is that although the exact version of HHH has low global sensitivity (as counting queries are 1-sensitive), the approximation functions due to streaming have high global sensitivity, linear in the available space. Despite this obstacle, we show that the absolute error for estimating frequencies in the streaming setting is independent of the available space.
- 4 November 2024
- Published by Association for Computing Machinery (ACM) in Proceedings of the ACM on Management of Data
- Vol. 2 (5) , 1-25
Open AccessDifferentially Private Hierarchical Heavy Hitters
The task of finding Hierarchical Heavy Hitters (HHH) was introduced by Cormode et al. [12] as a generalisation of the heavy hitter problem. While finding HHH in data streams has been studied extensively, the question of releasing HHH when the underlying data is private remains unexplored. In this paper, we formalise and study the notion of differentially private HHH, in both the streaming and non-streaming setting. In the non-streaming setting, we show the surprising result that the relative error in estimating the count for any prefix is independent of the height of the hierarchy and the number of heavy hitters in the stream. Additionally, our algorithms also improve the error guarantees of Ghazi et al. [24] for the problem of counting over trees. Meanwhile, in the streaming setting, the main issue is that although the exact version of HHH has low global sensitivity (as counting queries are 1-sensitive), the approximation functions due to streaming have high global sensitivity, linear in the available space. Despite this obstacle, we show that the absolute error for estimating frequencies in the streaming setting is independent of the available space.
- Open Access
- 29 October 2024
- Published by Association for Computing Machinery (ACM)
- p. 477-480
A Geospatial Perspective on Data Ownership, the Right to be Forgotten, Copyrights, and Plagiarism in Generative AI
Ethical use of data in generative AI is a growing concern. Large generative AI models are trained on pervasive data sets from a variety of sources, where the data records are fused into the models and become inseparable from them. This raises questions regarding data ownership and the use of personal data, artwork, and copyrighted content in generative AI models. Should people and businesses be allowed to request the removal of their data from a model, even if the data were collected in public places? Should the use of data in generative AI vary across different places based on local copyright laws? How should local laws and regulations regarding data misuse and harmful content be enforced? Can people and organizations verify that their data records have been removed from models or are being used properly? In this paper we discuss the geospatial aspects of data ownership in large generative AI models. We present a vision of generative AI applications that are aware of data ownership and location provenance, based on spatio-temporal features of the data and the usage. These aspects of location-aware AI governance could mitigate some of the risks associated with generative AI and support ethical use of it, in both local and global applications.
- 29 October 2024
- Published by Association for Computing Machinery (ACM)
- p. 477-480
Open AccessA Geospatial Perspective on Data Ownership, the Right to be Forgotten, Copyrights, and Plagiarism in Generative AI
Ethical use of data in generative AI is a growing concern. Large generative AI models are trained on pervasive data sets from a variety of sources, where the data records are fused into the models and become inseparable from them. This raises questions regarding data ownership and the use of personal data, artwork, and copyrighted content in generative AI models. Should people and businesses be allowed to request the removal of their data from a model, even if the data were collected in public places? Should the use of data in generative AI vary across different places based on local copyright laws? How should local laws and regulations regarding data misuse and harmful content be enforced? Can people and organizations verify that their data records have been removed from models or are being used properly? In this paper we discuss the geospatial aspects of data ownership in large generative AI models. We present a vision of generative AI applications that are aware of data ownership and location provenance, based on spatio-temporal features of the data and the usage. These aspects of location-aware AI governance could mitigate some of the risks associated with generative AI and support ethical use of it, in both local and global applications.
- Open Access
- 29 October 2024
- Published by Association for Computing Machinery (ACM)
- p. 661-664
Simulating Diffraction by Ray Tracing for Modeling 5G Networks
When planning cellular networks, the goal is to position antennas and adjust their parameters to maximize the coverage and minimize the interference between antennas. In new generations of cellular networks like 5G, network planning is becoming critical due to the use of high frequencies and the densification of the network. Antenna locations are typically decided based on the availability of cellular towers, however, tunable parameters, like tilt and transmission power, determine the network capacity and the quality of service. To optimize the network and compute coverage and interference for different tilt and power values, radio propagation is simulated using ray tracing over a geospatial model of the environment. In this paper, we present a ray-tracing module that computes long-distance effects of cellular transmissions, to accurately model interference between remote antennas. The two main novelties are (1) the computation of diffraction to include non-line-of-sight propagation, and (2) parallel computation for efficiency and scalability. The demonstration presents the geospatial effect of diffraction on the computation and the use of GPUs for scalability while avoiding race condition in the transformation from a polar coordinate system to a Cartesian representation.
- 29 October 2024
- Published by Association for Computing Machinery (ACM)
- p. 661-664
Open AccessSimulating Diffraction by Ray Tracing for Modeling 5G Networks
When planning cellular networks, the goal is to position antennas and adjust their parameters to maximize the coverage and minimize the interference between antennas. In new generations of cellular networks like 5G, network planning is becoming critical due to the use of high frequencies and the densification of the network. Antenna locations are typically decided based on the availability of cellular towers, however, tunable parameters, like tilt and transmission power, determine the network capacity and the quality of service. To optimize the network and compute coverage and interference for different tilt and power values, radio propagation is simulated using ray tracing over a geospatial model of the environment. In this paper, we present a ray-tracing module that computes long-distance effects of cellular transmissions, to accurately model interference between remote antennas. The two main novelties are (1) the computation of diffraction to include non-line-of-sight propagation, and (2) parallel computation for efficiency and scalability. The demonstration presents the geospatial effect of diffraction on the computation and the use of GPUs for scalability while avoiding race condition in the transformation from a polar coordinate system to a Cartesian representation.
- Open Access
- 13 November 2023
- Published by Association for Computing Machinery (ACM)
- p. 41-50
Cellular Network Optimization by Deep Reinforcement Learning and AI-Enhanced Ray Tracing
In this paper we study the use of deep reinforcement learning that is supported by a ray tracer, on top of a detailed 3D model of the geospatial environment, for optimization of antenna tilts in cellular networks. We propose two novel mechanisms---geospatial importance sampling and multi-path coefficient---to efficiently pass geospatial information to the reinforcement learning model. We show that this approach can be used for fast and scalable optimization of tilt levels of cellular antennas. We present an experimental evaluation that compares the use of reinforcement learning to greedy search, simulated annealing and Bayesian optimization. Our study shows that reinforcement learning is effective and can cope with optimization problems that are at a greater scale than the settings the other algorithms can cope with.
- 13 November 2023
- Published by Association for Computing Machinery (ACM)
- p. 41-50
Open AccessCellular Network Optimization by Deep Reinforcement Learning and AI-Enhanced Ray Tracing
In this paper we study the use of deep reinforcement learning that is supported by a ray tracer, on top of a detailed 3D model of the geospatial environment, for optimization of antenna tilts in cellular networks. We propose two novel mechanisms---geospatial importance sampling and multi-path coefficient---to efficiently pass geospatial information to the reinforcement learning model. We show that this approach can be used for fast and scalable optimization of tilt levels of cellular antennas. We present an experimental evaluation that compares the use of reinforcement learning to greedy search, simulated annealing and Bayesian optimization. Our study shows that reinforcement learning is effective and can cope with optimization problems that are at a greater scale than the settings the other algorithms can cope with.
- Open Access
- 12 January 2023
- Published by Association for Computing Machinery (ACM) in ACM Transactions on Spatial Algorithms and Systems
- Vol. 9 (1) , 1-30
Planning Wireless Backhaul Links by Testing Line of Sight and Fresnel Zone Clearance
Microwave backhaul links are often used as wireless connections between telecommunication towers, in places where deploying optical fibers is impossible or too expensive. The relatively high frequency of microwaves increases their ability to transfer information at a high rate, but it also makes them susceptible to spatial obstructions and interference. Hence, when deploying wireless links, there are two conflicting considerations. First, the antennas height, selected from the available slots on each tower, should be as low as possible. Second, there should be a line of sight (LoS) between the antennas, and a buffer around the LoS defined by the first Fresnel zone should be clear of obstacles. To compute antenna heights, a planning system for wireless links has to maintain an elevation model, efficiently discover obstacles between towers, and execute Fresnel-zone clearance tests over a 3D model of the deployment area. In this article we present a system and algorithms for computing the height of antennas, by testing LoS and clearance of Fresnel zones. The system handles the following requirements: (1) the need to cover large areas, e.g., all of the USA, (2) big distance between towers, e.g., 100 kilometers, and (3) computing batches of thousands of pairs within a few minutes. We introduce three novel algorithms for efficient computation of antenna heights, we show how to effectively model and manage the large-scale geospatial data needed for the planning, and we present the results of tests over real-world settings.
- 12 January 2023
- Published by Association for Computing Machinery (ACM) in ACM Transactions on Spatial Algorithms and Systems
- Vol. 9 (1) , 1-30
Open AccessPlanning Wireless Backhaul Links by Testing Line of Sight and Fresnel Zone Clearance
Microwave backhaul links are often used as wireless connections between telecommunication towers, in places where deploying optical fibers is impossible or too expensive. The relatively high frequency of microwaves increases their ability to transfer information at a high rate, but it also makes them susceptible to spatial obstructions and interference. Hence, when deploying wireless links, there are two conflicting considerations. First, the antennas height, selected from the available slots on each tower, should be as low as possible. Second, there should be a line of sight (LoS) between the antennas, and a buffer around the LoS defined by the first Fresnel zone should be clear of obstacles. To compute antenna heights, a planning system for wireless links has to maintain an elevation model, efficiently discover obstacles between towers, and execute Fresnel-zone clearance tests over a 3D model of the deployment area. In this article we present a system and algorithms for computing the height of antennas, by testing LoS and clearance of Fresnel zones. The system handles the following requirements: (1) the need to cover large areas, e.g., all of the USA, (2) big distance between towers, e.g., 100 kilometers, and (3) computing batches of thousands of pairs within a few minutes. We introduce three novel algorithms for efficient computation of antenna heights, we show how to effectively model and manage the large-scale geospatial data needed for the planning, and we present the results of tests over real-world settings.
- 25 March 2017
- Published by Springer Nature in GeoInformatica
- Vol. 22 (3) , 615-660
Combined geo-social search: computing top-k join queries over incomplete information
- 25 March 2017
- Published by Springer Nature in GeoInformatica
- Vol. 22 (3) , 615-660
Combined geo-social search: computing top-k join queries over incomplete information