1. Identification of women with high grade histopathology results after conisation by artificial neural networksMarko Mlinarič, Miljenko Križmarić, Iztok Takač, Alenka Repše-Fokter, 2022, izvirni znanstveni članek Povzetek: Background: The aim of the study was to evaluate if artificial neural networks can predict high-grade histopathology results after conisation from risk factors and their combinations in patients undergoing conisation because of pathological changes on uterine cervix. Patients and methods: We analysed 1475 patients who had conisation surgery at the University Clinic for Gynaecology and Obstetrics of University Clinical Centre Maribor from 1993-2005. The database in different datasets was arranged to deal with unbalance data and enhance classification performance. Weka open-source software was used for analysis with artificial neural networks. Last Papanicolaou smear (PAP) and risk factors for development of cervical dysplasia and carcinoma were used as input and high-grade dysplasia Yes/No as output result. 10-fold cross validation was used for defining training and holdout set for analysis. Results: Baseline classification and multiple runs of artificial neural network on various risk factors settings were performed. We achieved 84.19% correct classifications, area under the curve 0.87, kappa 0.64, F-measure 0.884 and Matthews correlation coefficient (MCC) 0.640 in model, where baseline prediction was 69.79%. Conclusions: With artificial neural networks we were able to identify more patients who developed high-grade squamous intraepithelial lesion on final histopathology result of conisation as with baseline prediction. But, characteristics of 1475 patients who had conisation in years 1993-2005 at the University Clinical Centre Maribor did not allow reliable prediction with artificial neural networks for every-day clinical practice.
Ključne besede: artificial neural networks, conisation, uterine cervical cancer, uterine cervical dysplasia, displazija materničnega vratu, rak materničnega vratu, konizacija, umetne nevronske mreže
Objavljeno v DiRROS: 24.07.2024; Ogledov: 84; Prenosov: 51
 Celotno besedilo (663,31 KB)
Gradivo ima več datotek! Več... |
2. Learning deep representations of enzyme thermal adaptationGang Li, Filip Buric, Jan Zrimec, Sandra Viknander, Jens Nielsen, Aleksej Zelezniak, Martin K. M. Engqvist, 2022, izvirni znanstveni članek Povzetek: Temperature is a fundamental environmental factor that shapes the evolution of organisms. Learning thermal determinants of protein sequences in evolution thus has profound significance for basic biology, drug discovery, and protein engineering. Here, we use a data set of over 3 million BRENDA enzymes labeled with optimal growth temperatures (OGTs) of their source organisms to train a deep neural network model (DeepET). The protein-temperature representations learned by DeepET provide a temperature-related statistical summary of protein sequences and capture structural properties that affect thermal stability. For prediction of enzyme optimal catalytic temperatures and protein melting temperatures via a transfer learning approach, our DeepET model outperforms classical regression models trained on rationally designed features and other deep-learning-based representations. DeepET thus holds promise for understanding enzyme thermal adaptation and guiding the engineering of thermostable enzymes.
Ključne besede: bioinformatics, deep neural networks, enzyme catalytic temperatures, optimal growth temperatures, protein thermostability, transfer learning
Objavljeno v DiRROS: 17.07.2024; Ogledov: 117; Prenosov: 108
Celotno besedilo (2,61 MB)
Gradivo ima več datotek! Več... |
3. Towards deploying highly quantized neural networks on FPGA using chiselJure Vreča, Anton Biasizzo, 2023, objavljeni znanstveni prispevek na konferenci Povzetek: We present chisel4ml, a Chisel-based tool that generates hardware for highly quantized neural networks described in QKeras. Such networks typically use parameters with bitwidths less than 8 bits and may have pruned connections. Chisel4ml can generate the highly quantized neural network as a single combinational circuit with pipeline registers in between the different layers. It supports heterogeneous quantization where each layer can have a different precision. The full parallelization enables very low-latency and high throughput inference, that are required for certain tasks. We illustrate this on the triggering system for the CERN Large Hadron Collider, which filters out events of interest and sends them on for further processing. We compare our tool against hls4ml, a high-level synthesis based approach for deploying similar neural networks. Chisel4ml is still under development. However, it already achieves comparable results to hls4ml for some neural network architectures. Chisel4ml is available on https://github.com/cs-jsi/chisel4ml.
Ključne besede: neural networks, QKeras, Chisel4ml
Objavljeno v DiRROS: 23.04.2024; Ogledov: 218; Prenosov: 131
Celotno besedilo (419,83 KB)
Gradivo ima več datotek! Več... |
4. |
5. Modelling seasonal dynamics of secondary growth in RJernej Jevšenak, Jožica Gričar, Sergio Rossi, Peter Prislan, 2022, izvirni znanstveni članek Povzetek: The monitoring of seasonal radial growth of woody plants addresses the ultimate question of when, how and why trees grow. Assessing the growth dynamics is important to quantify the effect of environmental drivers and understand how woody species will deal with the ongoing climatic changes. One of the crucial steps in the analyses of seasonal radial growth is to model the dynamics of xylem and phloem formation based on increment measurements on samples taken at relatively short intervals during the growing season. The most common approach is the use of the Gompertz equation, while other approaches, such as general additive models (GAMs) and generalised linear models (GLMs), have also been tested in recent years. For the first time, we explored artificial neural networks with Bayesian regularisation algorithm (BRNNs) and show that this method is easy to use, resistant to overfitting, tends to yield s-shaped curves and is therefore suitable for deriving temporal dynamics of secondary tree growth. We propose two data processing algorithms that allow more flexible fits. The main result of our work is the XPSgrowth() function implemented in the radial Tree Growth (rTG) R package, that can be used to evaluate and compare three modelling approaches: BRNN, GAM and the Gompertz function. The newly developed function, tested on intra-seasonal xylem and phloem formation data, has potential applications in many ecological and environmental disciplines where growth is expressed as a function of time. Different approaches were evaluated in terms of prediction error, while fitted curves were visually compared to derive their main characteristics. Our results suggest that there is no single best fitting method, therefore we recommend testing different fitting methods and selection of the optimal one.
Ključne besede: artificial neural networks, cambium, generalized additive model, Gompertz function, growing season, intra-annual time series
Objavljeno v DiRROS: 21.07.2022; Ogledov: 657; Prenosov: 411
Celotno besedilo (1,26 MB)
Gradivo ima več datotek! Več... |
