Image captioning is an important task for improving human-computer interaction as well as for a deeper understanding of the mechanisms underlying the image description by human. In recent years, this research field has rapidly developed and a number of impressive results have been achieved. The typical models are based on a neural networks, including convolutional ones for encoding images and recurrent ones for decoding them into text. More than that, attention mechanism and transformers are actively used for boosting performance. However, even the best models have a limit in their quality with a lack of data. In order to generate a variety of descriptions of objects in different situations you need a large training set. The current commonly used datasets although rather large in terms of number of images are quite small in terms of the number of different captions per one image. We expanded the training dataset using text augmentation methods. Methods include augmentation with synonyms as a baseline and the state-of-the-art language model called Bidirectional Encoder Representations from Transformers (BERT). As a result, models that were trained on a datasets augmented show better results than that models trained on a dataset without augmentation.

No abstract

No abstract

No abstract

Due to dramatic progress in high-throughput sequencing technologies and widespread of microarray assays over the last decade, gene expression data has been accumulating at an accelerating pace. All this insured gene expression profiling to be extensively used as a powerful technique for phenotype classification in many biological studies. However, this is not always possible to replicate a particular experiment with various organisms or tissues to achieve sample size that will be large enough to meet the assumptions of classical statistical methods used to deliver reliable classification results. Small dataset size due to lack of sample objects can also be a problem when trying to reuse the data from public databases submitted by other researchers from their experiments. In this paper we introduce a two-step classification method for a specific task of phenotype identification, which firstly clusters data and then performs classification within each cluster. We apply this method to a real dataset for the purpose of bacterial gene-expression analysis and present its results.