Multivariate Time Series Information Bottleneck

Time series (TS) and multiple time series (MTS) predictions have historically paved the way for distinct families of deep learning models. The temporal dimension, distinguished by its evolutionary sequential aspect, is usually modeled by decomposition into the trio of "trend, seasonality, noise", by attempts to copy the functioning of human synapses, and more recently, by transformer models with self-attention on the temporal dimension. These models may find applications in finance and e-commerce, where any increase in performance of less than 1% has large monetary repercussions, they also have potential applications in natural language processing (NLP), medicine, and physics. To the best of our knowledge, the information bottleneck (IB) framework has not received significant attention in the context of TS or MTS analyses. One can demonstrate that a compression of the temporal dimension is key in the context of MTS. We propose a new approach with partial convolution, where a time sequence is encoded into a two-dimensional representation resembling images. Accordingly, we use the recent advances made in image extension to predict an unseen part of an image from a given one. We show that our model compares well with traditional TS models, has information-theoretical foundations, and can be easily extended to more dimensions than only time and space. An evaluation of our multiple time series-information bottleneck (MTS-IB) model proves its efficiency in electricity production, road traffic, and astronomical data representing solar activity, as recorded by NASA's interface region imaging spectrograph (IRIS) satellite.

Analysis of the Future Evolution of Biocapacity and Landscape Characteristics in the Agro-Pastoral Zone of Northern China

The Agro-Pastoral Zone of Northern China (AZNC) is an ecologically fragile zone. It is a challenge to create scientifically sound plans for environmental conservation and agro-pastoral development due to the lack of future evolution prediction, and analysis of biocapacity (BC) and landscape characteristics. Using the Globeland30 dataset from 2000 to 2020, this study simulated 2030 land use/land cover (LULC) scenarios, and analyzed the future evolution of BC and landscape patterns. The results show that: (1) The Logistic and CA-Markov models can reasonably simulate the LULC changes in the research area, with ROC indices over 0.9 and Kappa approaching 0.805, after considering the driving factors such as physical geography, regional climate, and socio-economic development. (2) From 2000 to 2030, the spatial distribution pattern of LULC does not change significantly, and cultivated land, grassland, and forest are still the dominant land types in the research area. The regional BC exhibits an increasing trend (+4.55 × 10⁶ gha/a), and the spatial distribution pattern of BC is similar to that of LULC. (3) Changes in land miniaturization, landscape fragmentation, and decreased aggregation can be seen in the entire AZNC and specific land categories, including cultivated land, grassland, and forest. The study provides suggestions for formulating the AZNC's future ecological protection and agro-pastoral development strategies, and guidance for the LULC simulation in other agro-pastoral zones.

[PREDICTIVE MODEL OF THE PROBABLE CONTRACTION OF PSYCHOSOMATIC DISORDERS AMONG THE EXTREME PROFILE SPECIALISTS, A CASE STUDY OF SPECIALISTS IN UTILIZATION OF RADIOACTIVE SUBSTANCES]

The aim of the study is to develop a predictive model of the probable contraction of psychosomatic disorders among specialists in the utilization of radioactive substances. The predictive model was developed in order to prevent occupational diseases and preserve the mental health of specialists. The development of the model was carried out on the basis of the results of empirical research focused on identifying biopsychosocial factors that determine or, on the contrary, block the development of vitality in specialists. The vitality is here understood by us as the ability to manage the internal personal resources in the professional situations of personal stress, thereby ensuring one`s professional health. 179 specialists in the utilization of radioactive substances were examined. The study was carried out using a set of specially selected techniques: Scale «Psychosomatic disorders» by V. V. Boyko, methods for studying psychophysiological status (data on the brain wave activity (EEG); assessment of the functional state of the nervous system based on a simple and complex visual motor reaction; the color test by M. Luscher), individual psychological characteristics (the method of colour metaphors by I. L. Solomin, diagnostics of latent motivation, the Mini-Mult questionnaire, «The dominant type of interpersonal relations» methodology) and social situation of life (the questionnaire by the authors). As a result of the study, there were identified 32 features that have a statistically significant relationship with indicators of psychosomatic disorders (p < 0.001). The identified features were included in multiple logistic regression analysis as independent variables. As a result of further processing of the empirical data, carried out with the help of the method of step-by-step exclusion of variables, a predictive model of the likelihood of psychosomatic disorders in radioactive waste utolozation specialists was constructed, which was based on a subjective assessment of the living standards, the presence of a distrustful-skeptical type in interpersonal relationships and manifestations of paranoia.