A5085237554- Neural Networks and Applications
- Music and Audio Processing
- Bioinformatics and Genomic Networks
- Gene Regulatory Network Analysis
- Gene expression and cancer classification
- Computational Geometry and Mesh Generation
- Speech Recognition and Synthesis
- Data Management and Algorithms
- Advanced Neural Network Applications
- Machine Learning in Bioinformatics
- Robotic Path Planning Algorithms
- Data Visualization and Analytics
- Computational Physics and Python Applications
- Ferroelectric and Negative Capacitance Devices
- Bacterial Genetics and Biotechnology
- Natural Language Processing Techniques
- Formal Methods in Verification
- Interconnection Networks and Systems
- Image Retrieval and Classification Techniques
- Adversarial Robustness in Machine Learning
- Video Surveillance and Tracking Methods
- VLSI and FPGA Design Techniques
- Metaheuristic Optimization Algorithms Research
- Graph Theory and Algorithms
- Complexity and Algorithms in Graphs
Institute of Electronics and Computer Science
2013-2025
University of Latvia
2003-2022
Abstract Motivation: Microarray experiments comparing expression levels of all genes in yeast for hundreds mutants allow us to examine properties gene regulatory networks on a genomic scale. We can investigate questions such as network modularity, connectivity, and look with particular roles the structure. Results: have built genome-wide disruption yeast, using representation data directed labelled graphs. Nodes represent arcs connect nodes if source significantly alters target gene. are...
The study focuses on creating symbol-shaped flame designs by optimizing dynamics, using machine learning techniques integrated with advanced combustion modelling. This research addresses the limitations of traditional pyrotechnic systems, which lack precise control over shapes, introducing a method that uses differentiable physics simulations for shaping flames. optimization framework PhiFlow and includes radiative heat loss chemical kinetics improved accuracy. Machine is used not only to...
Modern neural networks obtain information about the problem and calculate output solely from input values. We argue that it is not always optimal, network's performance can be significantly improved by augmenting with a query mechanism allows network at run time to make several solution trials get feedback on loss value each trial. To demonstrate capabilities of mechanism, we formulate an unsupervised (not depending labels) function for Boolean Satisfiability Problem (SAT) theoretically show...
Mass spectrometry (MS)‐based quantitative proteomics experiments typically assay a subset of up to 60% the ≈20 000 human protein coding genes. Computational methods for imputing missing values using RNA expression data usually allow only imputations proteins measured in at least some samples. In silico comprehensively estimating abundances across all are still missing. Here, novel method is proposed deep learning extrapolate observed label‐free MS proteins, leveraging gene functional...
A key requirement in sequence to processing is the modeling of long range dependencies. To this end, a vast majority state-of-the-art models use attention mechanism which O($n^2$) complexity that leads slow execution for sequences. We introduce new Shuffle-Exchange neural network model tasks have O(log n) depth and O(n log total complexity. show powerful enough infer efficient algorithms common algorithmic benchmarks including sorting, addition multiplication. evaluate our architecture on...
Parcel sorting is becoming a significant challenge for delivery distribution centers and mostly automated by using high-throughput machinery, but manual work still used to feed these machines placing the parcels on conveyor belt. In this paper, an AI-based robotic solution that automates parcel placement task was developed. The architecture of proposed system along with methods how implement it are described currently available hardware software components. choices lead well-functioning...
A video processing algorithm for vehicle parameter acquisition and classification is presented. The based on combination of several detection lines. According to passing vehicles, intervals are created the Intervals different lines, belonging same vehicle, combined. Further allows acquire parameters classify vehicles. accuracy counting analyzed videos.
In this paper we address the problem of visualizing overlapping sets points with a fixed positioning in comprehensible way. A standard visualization technique is to enclose point isocontours generated by bounding potential field function. The most commonly used functions are various approximations Gaussian distribution. Such an approach produces smooth and appealing shapes, however it may produce incorrect nesting regions, e.g. some contained inside foreign set region. We introduce different...
Attention is a commonly used mechanism in sequence processing, but it of O(n^2) complexity which prevents its application to long sequences. The recently introduced neural Shuffle-Exchange network offers computation-efficient alternative, enabling the modelling long-range dependencies O(n log n) time. model, however, quite complex, involving sophisticated gating derived from Gated Recurrent Unit. In this paper, we present simple and lightweight variant network, based on residual employing...
Algorithm learning is a core problem in artificial intelligence with significant implications on automation level that can be achieved by machines. Recently deep methods are emerging for synthesizing an algorithm from its input-output examples, the most successful being Neural GPU, capable of multiplication. We present several improvements to GPU substantially reduces training time and improves generalization. introduce new technique - hard nonlinearities saturation costs- has general...
In this paper we present a study of logo detection in images from media agency. We compare two most widely used methods - HOG and SIFT on challenging dataset arising printed press news portals. Despite common opinion that method is superior, our results show performs significantly better dataset. augment the with image resizing rotation to improve its performance even more. found out by using such approach it possible obtain good increased recall reasonably decreased precision.
Abstract Background Current Hi-C technologies for chromosome conformation capture allow to understand a broad spectrum of functional interactions between genome elements. Although significant progress has been made into analysis data identify biologically features, many questions still remain open, in particular regarding potential biological significance various topological features that are characteristic chromatin interaction networks. Results It previously observed promoter (PCHi-C)...
Current high-throughput experimental techniques make it feasible to infer gene regulatory interactions at the whole-genome level with reasonably good accuracy. Such experimentally inferred networks have become available for a number of simpler model organisms such as S. cerevisiae, and others. The availability provides an opportunity compare processes whole genome level, in particular, assess similarity homologous pairs either from same or different species. We present here new technique...