Yan Mayster - Academia.edu (original) (raw)

Papers by Yan Mayster

Modern telemetry entails the gathering of data (per location or time interval) in such large quan... more Modern telemetry entails the gathering of data (per location or time interval) in such large quantities that the data often exceeds the capacity of the communication channels to a server (or cloud). An example of such telemetry is the gathering of street-level imagery and video where, even when the data-collecting vehicle has a wireless 4G connection, a very large amount of data is gathered so quickly that it is written on storage media such as solid state, optical, magnetic storage, etc. and physically mailed to a processing center. Besides the delay attributable to physical mailing, manual processing of the storage media at the point of capture also causes a substantial delay. This disclosure describes techniques to optimize the latency and bandwidth of sensoracquired telemetry data by rapidly determining, e.g., at the point of capture, the value of specific pieces of data (image, sequence of video frames, etc.), and by utilizing the available 2 Bahnsen and Mayster: Optimized Upload of Telemetry Data

There are situations where people want to find geographic locations or areas that are similar to ... more There are situations where people want to find geographic locations or areas that are similar to a given area, e.g., in terms of its urban or natural features, types of buildings, parks, views, suitability for particular lifestyles or purposes, etc. This disclosure utilizes machine learning techniques to provide answers to the general question "where can I experience life 2 Mayster and Shucker: Multi-resolution Semantic Area Search

Springer eBooks, Apr 3, 2008

We consider the problem of weighted rectilinear approximation on the plane and offer both exact a... more We consider the problem of weighted rectilinear approximation on the plane and offer both exact algorithms and heuristics with provable performance bounds. Let S = {(p i , w i)} be a set of n points pi in the plane, with associated distance-modifying weights wi > 0. We present algorithms for finding the best fit to S among x-monotone rectilinear polylines R with a given number k < n of horizontal segments. We measure the quality of the fit by the greatest weighted vertical distance, i.e., the approximation error is max 1≤i≤n w i d v (p i , R), where d v (p i , R) is the vertical distance from pi to R. We can solve for arbitrary k optimally in O(n 2) or approximately in O(n log 2 n) time. We also describe a randomized algorithm with an O(n log 2 n) expected running time for the unweighted case and describe how to modify it to handle the weighted case in O(n log 3 n) expected time. All algorithms require O(n) space.

This thesis addresses several problems in the facility location sub-area of computational geometr... more This thesis addresses several problems in the facility location sub-area of computational geometry. Let S be a set of n points in the plane. We derive algorithms for approximating S by a step function curve of size k < n, i.e., by an x-monotone orthogonal polyline ℜ with k < n horizontal segments. We use the vertical distance to measure the quality of the approximation, i.e., the maximum distance from a point in S to the horizontal segment directly above or below it. We consider two types of problems: min-ε, where the goal is to minimize the error for a given number of horizontal segments k and min-#, where the goal is to minimize the number of segments for a given allowed error ε. After O(n) preprocessing time, we solve instances of the latter in O(min{k log n, n}) time per instance. We can then solve the former problem in O(min{n 2 , nk log n}) time. Both algorithms require O(n) space. The second contribution is a heuristic for the min-ε problem that computes a solution within a factor of 3 of the optimal error for k segments, or with at most the same error as the k-optimal but using 2k-1 segments. Furthermore, experiments on real data show even better results than what is guaranteed by the theoretical bounds. Both approximations run in O(n log n) time and O(n) space. Then, we present an exact algorithm for the weighted version of this problem that runs in O(n 2) time and generalize the heuristic to handle weights at the expense of an additional log n factor. At this point, a randomized algorithm that runs in O(n log 2 n) expected time for the unweighted version is presented. It easily generalizes to the weighted case, though at the expense of an additional log n factor. Finally, we treat the maximin problem and present an O(n 3 log n) solution to the problem of finding the furthest separating line through a set of weighted points. We conclude with solutions to the "obnoxious" wedge problem: an O(n 2 log n) algorithm for the general case of a wedge with its apex on the boundary of the convex hull of S and an O(n 2) algorithm for the case of the apex of a wedge coming from the input set S.

A common type of single-vehicle accident involves a vehicle entering a passageway with insufficie... more A common type of single-vehicle accident involves a vehicle entering a passageway with insufficient clearance. Variable vehicle dimensions, e.g., oversized trucks, automobiles with roof-mounted bicycles, etc. pose a special challenge since the dimensions of the vehicle are nonstandard for just the duration of the trip. Little is done today to alert drivers of impending

Large open spaces, e.g., beaches, cattle-grazing grounds, etc., often need to be monitored for se... more Large open spaces, e.g., beaches, cattle-grazing grounds, etc., often need to be monitored for security or other reasons. In many cases, monitoring from the air provides the widest and most effective coverage. However, due to the cost of air-based monitoring, such aerial monitoring is often not possible, putting individuals or assets at risk or causing loss of revenue. This disclosure describes techniques to deploy small aircraft, e.g., kites, blimps, low-powered drones, etc. to monitor a given area. A video feed from the aircraft is analyzed using a trained 2 Mayster and Shucker: Airship-based security and monitoring with machine learning

Existing approaches to adapt advertising messages on vehicle-mounted displays based on location r... more Existing approaches to adapt advertising messages on vehicle-mounted displays based on location rely on static demographic information about the area. Such approaches do not take into account changes in relevant contextual factors connected to the current location of the vehicle at a given time. This disclosure describes techniques to automatically select and display contextually relevant messages, such as advertisements, on vehicles. The messages are dynamically updated based on relevant changes in the vehicle’s context, such as location, time of day, day of the week, people in the vicinity, local events, current advertising needs of businesses, etc. If the route of a vehicle is known in advance fully or partially, the set of messages for the route can be pre-selected. When the route for a vehicle is flexible, the techniques can be utilized to select routes based on advertising needs

Apps that stream content often interleave the personalized content broadcast with other informati... more Apps that stream content often interleave the personalized content broadcast with other information, such as advertisements, announcements, etc. Personalized and contextually optimal selection of such additional content is currently constrained by the information available to the specific app provider. This disclosure describes a context aggregator service that app makers can query to enhance the selection of inserted content by querying a broad variety of relevant contextual information, obtained with user permission. The context aggregator service can utilize individual and/or aggregated information to provide query responses, e.g., topics or advertisements that a user is likely to be interested in. If the user permits, the context aggregator service can further determine advertising conversion based on subsequent user actions and use the conversion metrics in interest score calculations

This paper describes a system that would suggest parking choices to a user based on a destination... more This paper describes a system that would suggest parking choices to a user based on a destination selected and information previously extracted from regularly gathered aerial and/or street-level imagery. The system provides information on the expected occupancy of the suggested parking options at the estimated time of user arrival. Further, this system would be able to provide information on which parking options are free and which ones are commercial and the pricing information, also sensitive to the time of the user arrival. The occupancy is predicted via the application of machine learning models to the regularly gathered imagery. The imagery can be analyzed with machine learning models specialized to look for ground level parking lots with or without vehicles in them to find the locations of any open-air parking or multi-level parking lots with the roof level available for parking. Introduction Typical navigation systems do not provide navigation help to nearby parking lots or c...

There are provided systems and methods for ordering an image capture system. In one embodiment, a... more There are provided systems and methods for ordering an image capture system. In one embodiment, a method includes obtaining data describing a plurality of images that are associated with a geographic area. The method includes analyzing at least a subset of the plurality of images to determine an occurrence of one or more changes associated with the geographical area. The method includes determining a geographic area associated with the degree of change based at least in part on the occurrence of the one or more changes associated with the geographical area. The method includes providing a control command to an image capture system, in order at least partially to adjust a capturing image data that are associated with the geographic area based on the degree of change.

This thesis addresses several problems in the facility location sub-area of computational geometr... more This thesis addresses several problems in the facility location sub-area of computational geometry. Let S be a set of n points in the plane. We derive algorithms for approximating S by a step function curve of size k < n, i.e., by an x-monotone orthogonal polyline R with k < n horizontal segments. We use the vertical distance to measure the quality of the approximation, i.e., the maximum distance from a point in S to the horizontal segment directly above or below it. We consider two types of problems: min-ε, where the goal is to minimize the error for a given number of horizontal segments k and min-#, where the goal is to minimize the number of segments for a given allowed error ε. After O(n) preprocessing time, we solve instances of the latter in O(min{k log n, n}) time per instance. We can then solve the former problem in O(min{n 2 , nk log n}) time. Both algorithms require O(n) space. The second contribution is a heuristic for the min-ε problem that computes a solution within a factor of 3 of the optimal error for k segments, or with at most the same error as the k-optimal but using 2k − 1 segments. Furthermore, experiments on real data show even better results than what is guaranteed by the theoretical bounds. Both approximations run in O(n log n) time and O(n) space. Then, we present an exact algorithm for the weighted version of this problem that runs in O(n 2) time and generalize the heuristic to handle weights at the expense of an additional log n factor. At this point, a randomized I thank my advisor Mario Lopez for his many years of close friendship and mentoring. His patience, advice, and consistent encouragement were absolutely invaluable to me in completing this dissertation and degree. I also thank my committee members, Alvaro Arias, Chris GauthierDickey, Scott Leutenegger, and Ronald DeLyser, for their time, advice, and comments. In addition, I would like to thank Drs. GauthierDickey and DeLyser for the many grammatical and presentation suggestions to my thesis. I thank again my advisor Mario Lopez for inviting me to continue my graduate school experience with doctoral work. I also want to thank the faculty at the departments of Computer Science and Mathematics whose courses I have taken at the University of Denver throughout all my degrees. I am grateful to all of you. Last, but not least, I wholeheartedly thank my graduate student colleagues and coauthors, Mohammed Al-Bow and Riquelmi Cardona, for their friendship, collaboration, conversations, and all the time we have spent together. I also thank Jeff Edgington for leading the way and showing me the path to graduation and freedom. All of you friends and colleagues, whose names are too many to mention here, have imparted your knowledge and wisdom to me, and I thank you all. Most importantly, I thank my parents, Boris and Rita Mayster, and my brother, Dmitriy Mayster, for their love and attention.

Lecture Notes in Computer Science

Lecture Notes in Computer Science, 2006

We derive algorithms for approximating a set S of n points in the plane by an x-monotone rectilin... more We derive algorithms for approximating a set S of n points in the plane by an x-monotone rectilinear polyline with k horizontal segments. The quality of the approximation is measured by the maximum distance from a point in S to the segment above or below it. We consider two types of problems: min-ε, where the goal is to minimize the

Journal of Visual Communication and Image Representation, 2006

Let S be a set of n points in the plane. We derive algorithms for approximating S by a step funct... more Let S be a set of n points in the plane. We derive algorithms for approximating S by a step function of size k < n, i.e., by an x-monotone rectilinear polyline R with k < n horizontal segments. We use the vertical distance to measure the quality of the approximation, i.e., the maximum distance from a point in S to the horizontal segment directly above or below it. We consider two types of problems: mine , where the goal is to minimize the error for a given number of horizontal segments k and min-#, where the goal is to minimize the number of segments for a given allowed error e. After O (n) preprocessing time, we solve instances of the latter in O (min{k log n, n}) time per instance. We can then solve the former problem in O (min{n 2 , nk log n}) time. Both algorithms require O (n) space. Our second contribution is an approximation algorithm for the mine problem that computes a solution within a factor of 3 of the optimal error for k segments, or with at most the same error as the k-optimal but using 2k À 1 segments. Furthermore, experiments on real data show even better results than what is guaranteed by the theoretical bounds. Both approximations run in O (n log n) time and O (n) space.

Springer eBooks, Apr 3, 2008

This thesis addresses several problems in the facility location sub-area of computational geometr... more This thesis addresses several problems in the facility location sub-area of computational geometry. Let S be a set of n points in the plane. We derive algorithms for approximating S by a step function curve of size k < n, i.e., by an x-monotone orthogonal polyline R with k < n horizontal segments. We use the vertical distance to measure the quality of the approximation, i.e., the maximum distance from a point in S to the horizontal segment directly above or below it. We consider two types of problems: min-ε, where the goal is to minimize the error for a given number of horizontal segments k and min-#, where the goal is to minimize the number of segments for a given allowed error ε. After O(n) preprocessing time, we solve instances of the latter in O(min{k log n, n}) time per instance. We can then solve the former problem in O(min{n 2 , nk log n}) time. Both algorithms require O(n) space. The second contribution is a heuristic for the min-ε problem that computes a solution within a factor of 3 of the optimal error for k segments, or with at most the same error as the k-optimal but using 2k − 1 segments. Furthermore, experiments on real data show even better results than what is guaranteed by the theoretical bounds. Both approximations run in O(n log n) time and O(n) space. Then, we present an exact algorithm for the weighted version of this problem that runs in O(n 2) time and generalize the heuristic to handle weights at the expense of an additional log n factor. At this point, a randomized I thank my advisor Mario Lopez for his many years of close friendship and mentoring. His patience, advice, and consistent encouragement were absolutely invaluable to me in completing this dissertation and degree. I also thank my committee members, Alvaro Arias, Chris GauthierDickey, Scott Leutenegger, and Ronald DeLyser, for their time, advice, and comments. In addition, I would like to thank Drs. GauthierDickey and DeLyser for the many grammatical and presentation suggestions to my thesis. I thank again my advisor Mario Lopez for inviting me to continue my graduate school experience with doctoral work. I also want to thank the faculty at the departments of Computer Science and Mathematics whose courses I have taken at the University of Denver throughout all my degrees. I am grateful to all of you. Last, but not least, I wholeheartedly thank my graduate student colleagues and coauthors, Mohammed Al-Bow and Riquelmi Cardona, for their friendship, collaboration, conversations, and all the time we have spent together. I also thank Jeff Edgington for leading the way and showing me the path to graduation and freedom. All of you friends and colleagues, whose names are too many to mention here, have imparted your knowledge and wisdom to me, and I thank you all. Most importantly, I thank my parents, Boris and Rita Mayster, and my brother, Dmitriy Mayster, for their love and attention.

Lecture Notes in Computer Science

Lecture Notes in Computer Science, 2006

Journal of Visual Communication and Image Representation, 2006