The text-to-SQL task is an active challenge in Natural Language Processing. Many existing solutions focus on using black-box language models extended with specialized components within customized end-to-end text-to-SQL pipelines. While these solutions use both closed-source proprietary language models and coding-oriented open-source models, there is a lack of research regarding SQL-specific small generative models. At the same time, recent advancements in self-correcting generation strategies show promise for improving the capabilities of existing architectures. The application of these concepts to the text-to-SQL task remains unexplored. In this paper, we introduce RetrySQL, a new approach to training text-to-SQL generation models. We prepare reasoning steps for reference SQL queries and then corrupt them to create retry data that contains both incorrect and corrected steps, divided with a special token. We continuously pre-train open-source coding models with this data and demonstrate that retry steps yield an improvements of up to 4 and 9 percentage points for overall and challenging execution metrics, respectively, as compared to pre-training without retry data. We showcase that the self-correcting behavior is learned by the model and the increase in downstream accuracy metrics is a result of this additional skill. Finally, we incorporate RetrySQL-trained models into the full text-to-SQL pipeline and showcase that they are competitive in terms of execution accuracy with proprietary models that contain orders of magnitude more parameters. RetrySQL demonstrates that self-correction can be learned in the text-to-SQL task and provides a novel way of improving generation accuracy for small SQL-oriented language models.

The study investigates the effectiveness of a differential neural network architecture in ultrasonic tomography (UST) for industrial applications. The proposed model employs a dual-branch structure, where each branch independently processes identical input data before passing the outputs to a differential layer. This approach enhances the model's ability to capture residual components, improving the reconstruction of tomographic images. Experiments were conducted using a tomographic system with 16 transducers, generating training and validation datasets. Comparative analysis between a differential LSTM-based network and a standard LSTM model demonstrated that the differential architecture achieved superior reconstruction quality. The results confirm that this approach enhances accuracy, sharpness, and overall image clarity, making it a promising solution for improving UST image reconstruction.

The article discusses research on a wearable medical diagnostic system using electrical impedance tomography. This system aims to diagnose long-term respiratory diseases, particularly COPD, ARDS, pneumothorax, pneumonia, bron-chospasm, and pulmonary hypertension. It seeks to reduce the number of tests needed for accurate diagnoses, thus saving time. The article compares two classification models for distinguishing between diseased and healthy individuals. This approach helps streamline the diagnostic process.

During the first stage of the project a computer vision system for human motion tracking was developed. For its implementation, a pair of cameras and a bicycle trainer are required. Human movement is monitored in real time by an effective algorithm that determines the key angles between the joints of a person exercising on a bike trainer. The following phase of the work focused on comparing the discussed system with a reference professional set, based on human motion sensors. The gathered data was then further analysed and compared, indicating that the accuracy of the developed system is fully satisfactory.

The paper presents a new method for building the CNN architecture with improved generalization capabilities. The network is formed from two deep structures differing by the activation functions. The dense layer signals are also separated into two branches and the verdicts of both are fused randomly into final decision using the classical cross-entropy loss formulation and softmax approach. In the learning process, the parameters of the locally connected CNN layers are adapted based on random combinations of the signals of these two parallel CNN subnetworks. The randomness introduced in all stages of the network structure fulfills the role of automatic implicit regularization at a very small number of learning data. The positive role of such regularization has been confirmed in numerical experiments of image recognition representing melanoma or non-melanoma cases. The results obtained by using the proposed method showed a significant improvement in numerical quality measures, including accuracy, sensitivity, F1 measure, and the area under the ROC curve.