“Evalds is a very talented developer, who has an exceptionally good grasp of the programming languages he employs, as well as how to systematically take a development project through design, to implementation, finally to deployment. Evalds is also good at recognising where other peoples skills lie and ensuring that development plans take into account peoples varying skillsets. He works well as part of a team, accepting the ideas, opinions and beliefs of others but never being afraid to express his own ideas, opinions and beliefs. He is a very hard worker and tends to encourage other people to work to his levels, bringing out the best in them.”
About
Working on Artificial Intelligence products that give people super-powers…
Activity
-
Imagine running production for a whole week and not noticing a critical defect. Surely, the production line operator should have noticed…
Imagine running production for a whole week and not noticing a critical defect. Surely, the production line operator should have noticed…
Liked by Evalds Urtans
-
At Starwatcher.io we are rolling out Country insights view. Besides having overall KPI's about country, it also allows to get a feel about startup…
At Starwatcher.io we are rolling out Country insights view. Besides having overall KPI's about country, it also allows to get a feel about startup…
Liked by Evalds Urtans
Experience
Education
-
Rigas Tehniska Universitate
-
Activities and Societies: Heading comunity of researchers of Data Science Latvia.
Research in Deep Reinforcement Learning, Deep Representation Learning and Loss functions.
-
- Present
Hadoop, Spark, Flink, Machine Learning
Startup & Business development -
-
Research in Active infrared markers for virtual and augmented reality is purposing synthesis of algorithms and methods to
create system that can recognize positions and orientations of markers that are invisible to humans.
It uses high frequency infrared camera of Leap Motion sensor and Arduino microcontroller that uses infrared light emitting diodes.
Positions and orientation acquired by the algorithm are display as overlays over physical
objects using Oculus Rift head mounted…Research in Active infrared markers for virtual and augmented reality is purposing synthesis of algorithms and methods to
create system that can recognize positions and orientations of markers that are invisible to humans.
It uses high frequency infrared camera of Leap Motion sensor and Arduino microcontroller that uses infrared light emitting diodes.
Positions and orientation acquired by the algorithm are display as overlays over physical
objects using Oculus Rift head mounted display.
It can be used in real time with Leap Motion's 111Hz frequency camera. Performance of system is limited only
by the speed of infrared camera and it would preform as well as traditional rectangular marker systems using 300Hz frequency infrared camera.
Purposed system can not only send identification of a marker, but also additional information like user interaction or data from sensors in object that has markers attached. -
-
AI algorithms for games
Graphics, shaders
Game design -
-
Publications
-
Exponential Triplet Loss
International Conference on Compute and Data Analysis, 2020, IEEE/ACM
This paper introduces a novel variant of the Triplet Loss function that converges faster and gives better results. This function can separate class instances homogeneously through the whole embedding space. With Exponential Triplet Loss function we also introduce a novel type of embedding space regularization Unit-Range and Unit-Bounce that utilizes euclidean space more efficiently and resembles features of the cosine distance. We also examined factors for choosing the best embedding vector…
This paper introduces a novel variant of the Triplet Loss function that converges faster and gives better results. This function can separate class instances homogeneously through the whole embedding space. With Exponential Triplet Loss function we also introduce a novel type of embedding space regularization Unit-Range and Unit-Bounce that utilizes euclidean space more efficiently and resembles features of the cosine distance. We also examined factors for choosing the best embedding vector size for specific embedding spaces. Finally, we also demonstrate how new function can train models for one-shot learning and re-identification tasks.
-
Value Iteration Solver Networks
International Conference on Intelligent Autonomous Systems, IEEE
Value Iteration Algorithm is iterative and can’t be parallelized. Computation time grows exponentially when the size of the input maps is increased. We propose UNet-RNN- Skip artificial neural network architecture that can be used to parallelize Value Iteration Algorithm results. The proposed model can solve Value Iteration problem in fewer iterations than the original algorithm and computation time increases by only a small amount when increasing the size of the input map. Fundamental…
Value Iteration Algorithm is iterative and can’t be parallelized. Computation time grows exponentially when the size of the input maps is increased. We propose UNet-RNN- Skip artificial neural network architecture that can be used to parallelize Value Iteration Algorithm results. The proposed model can solve Value Iteration problem in fewer iterations than the original algorithm and computation time increases by only a small amount when increasing the size of the input map. Fundamental UNet-RNN-Skip architecture can be used also to solve and parallelize other sequential problems. With this paper synthetic dataset of maps and generator has been published to enable further studies in mapping and path planning tasks.
-
Survey of Deep Q-Network variants in PyGame Learning Environment
ICDLT 2018, ACM
Q-value function models based on variations of Deep Q-Network (DQN) have shown good results in many virtual environments. In this paper, over 30 sub-algorithms were surveyed that influence the performance of DQN variants. Important stability and repeatability aspects of state of art Deep Reinforcement Learning algorithms were found. Multi Deep Q-Network (MDQN) as a generalization of popular Double Deep Q-Network (DDQN) algorithm was developed. Visual representations of a learning process as…
Q-value function models based on variations of Deep Q-Network (DQN) have shown good results in many virtual environments. In this paper, over 30 sub-algorithms were surveyed that influence the performance of DQN variants. Important stability and repeatability aspects of state of art Deep Reinforcement Learning algorithms were found. Multi Deep Q-Network (MDQN) as a generalization of popular Double Deep Q-Network (DDQN) algorithm was developed. Visual representations of a learning process as Q-Value maps were produced using PyGame Learning Environment. Videos of trained models available in following link
-
Software for Method Development in Liquid Chromatography
"Software-Assisted Method Development in High Performance Liquid Chromatography"; ISBN: 978-1-78634-545-5, Sep. 2018
-
ACTIVE INFRARED MARKERS FOR AUGMENTED AND VIRTUAL REALITY
Conference: 15th International Scientific Conference on Engineering for Rural Development 2016
Our research is proposing an algorithm and technical implementation of a system that can recognize positions and orientations of IR (Infrared) LED (Light Emitting Diode) markers that are invisible to a naked human eye. System is made to work with Oculus Rift DK2 head-mounted display coupled with a 111Hz IR Leap Motion camera. It adds functionality to these devices by allowing them to track different kinds of objects using active IR markers. Up to now most common way for tracking markers for…
Our research is proposing an algorithm and technical implementation of a system that can recognize positions and orientations of IR (Infrared) LED (Light Emitting Diode) markers that are invisible to a naked human eye. System is made to work with Oculus Rift DK2 head-mounted display coupled with a 111Hz IR Leap Motion camera. It adds functionality to these devices by allowing them to track different kinds of objects using active IR markers. Up to now most common way for tracking markers for augmented reality were using fiducial markers that are visible to human eye or other static markers. Such marker systems require flat surface for attaching sticker of a fiducial marker. With our system it is possible to create invisible markers and they can be attached to objects without flat surfaces. As a proof of concept we made a virtual ping-pong game where player uses a physical table tennis paddle fitted with active IR markers. At the time of research, no such tracking systems were available in public domain. Currently similar commercial systems are in development by Oculus and other companies. Limitations of hardware that we found in our research might be one of the main reasons why commercial product is not yet completed. In case of Oculus Rift's IR camera, it has insufficient 30Hz frame-rate to support active IR markers. We found that even with capture rate of 111Hz proposed system works 3 times slower than conventional fiducial markers, but starting at this frame-rate it can be applied for real-time applications.
Other authorsSee publication -
Association of Tumor Necrosis Factor-α (TNFα) Gene Polymorphisms with HLA Class II Alleles in HIV/AIDS Patients
British Journal of Medicine & Medical Research
Research of HLA class 2 genotypes of HIV patients. Machine learning algorithms were used for data mining.
Other authors -
Mobile robot path planning for indoor use
Conference: 13th International Scientific Conference on Engineering for Rural Development 2014
The paper presents practical and theoretical results acquired during a scientific project addressing multi-robot control and management challenges with possible application in greenhouse automation solutions. The main focus of the paper is path planning of mobile robots addressing computation time and memory constraints in embedded robotic systems. The paper presents an analysis that is based on practically experienced cases and proposes modifications of the discussed RRT-based methods in order…
The paper presents practical and theoretical results acquired during a scientific project addressing multi-robot control and management challenges with possible application in greenhouse automation solutions. The main focus of the paper is path planning of mobile robots addressing computation time and memory constraints in embedded robotic systems. The paper presents an analysis that is based on practically experienced cases and proposes modifications of the discussed RRT-based methods in order to ensure better quality of the planning result as well as saved time in specific cases. A key factor of the compareson analysis is time and memory usage that usually are limited in embedded devices like small scale mobile robots. The paper also presents experiment results collected using a prototype robotic system.
Other authorsSee publication
Projects
-
World Without Architect?
The first pilot project was exhibited in Tallinn Architecture Biennial, representing the power of collaboration between young architects and tech professionals. WWA? is provoking, experimenting how new technologies such as Virtual Reality, 3D printing and digitization transforms the matter of architects profession.
Other creatorsSee project -
Snowybirds
Snowy Birds is a VR-based video game, using Oculus Rift and Leap Motion tech for the ultimate snowball-fighting experience against the nasty birds.
Other creatorsSee project -
TV streaming / scheduling software
Multi-platform software running on Linux/Windows servers. Management trough web interface.
Output goes to digital TV provided by lattelecom.com -
Social Game - Hotel
Social game in portal http://frype.com/ with 133378 players online.
Other creatorsSee project -
www.baterijas.lv
Public contests system for schools.
Adaptations:
http://baterijas.lv/
http://skiropakas.lv/
http://zalajosta.asketic.comOther creatorsSee project -
Voice Synthesizer
-
Natural language synthesizer in Latvian from diphones.
-
www.lilla.lv
-
Online digital media/issue portal for women. Adaptive CSS, HTML5, JS & AJAX. Running on Wordpress. Integration of social networks & payment system.
Other creatorsSee project
Languages
-
English
-
-
Latvian
-
-
Russian
-
Recommendations received
1 person has recommended Evalds
Join now to viewMore activity by Evalds
-
Two things: 1. I found this awesome twitter account that only posts incredible utopian sci-fi landscapes combining mega-infrastructure with…
Two things: 1. I found this awesome twitter account that only posts incredible utopian sci-fi landscapes combining mega-infrastructure with…
Liked by Evalds Urtans
-
MAMY TO! BIELIK WYLĄDOWAŁ POWTARZAM, BIELIK WYLĄDOWAŁ
MAMY TO! BIELIK WYLĄDOWAŁ POWTARZAM, BIELIK WYLĄDOWAŁ
Liked by Evalds Urtans
-
Radość ogarnęła nas na wieść o kolaboracji z jednym z najlepszych polskich LLM, Bielik SpeakLeash.org! Z niecierpliwością oczekujemy na polskich…
Radość ogarnęła nas na wieść o kolaboracji z jednym z najlepszych polskich LLM, Bielik SpeakLeash.org! Z niecierpliwością oczekujemy na polskich…
Shared by Evalds Urtans
Other similar profiles
Explore collaborative articles
We’re unlocking community knowledge in a new way. Experts add insights directly into each article, started with the help of AI.
Explore MoreOthers named Evalds Urtans
2 others named Evalds Urtans are on LinkedIn
See others named Evalds Urtans