Multi-modal program inference: a marriage of pre-trained language models and component-based synthesis

Multi-modal program synthesis refers to the task of synthesizing programs (code) from their specification given in different forms, such as a combination of natural language and examples. Examples provide a precise but incomplete specification, and natural language provides an ambiguous but more "complete" task description. Machine-learned pre-trained models (PTMs) are adept at handling ambiguous natural language, but struggle with generating syntactically and semantically precise code. Program synthesis techniques can generate correct code, often even from incomplete but precise specifications, such as examples, but they are unable to work with the ambiguity of natural languages. We present an approach that combines PTMs with component-based synthesis (CBS): PTMs are used to generate candidates programs from the natural language description of the task, which are then used to guide the CBS procedure to find the program that matches the precise examples-based specification. We use our combination approach to instantiate multi-modal synthesis systems for two programming domains: the domain of regular expressions and the domain of CSS selectors. Our evaluation demonstrates the effectiveness of our domain-agnostic approach in comparison to a state-of-the-art specialized system, and the generality of our approach in providing multi-modal program synthesis from natural language and examples in different programming domains.

Modern fire safety systems in shopping and entertainment centers

Currently, the number of shopping centers in Russia is increasing. For their construction, new combustible materials and design solutions are used. This creates a high fire load and requires fire safety measures. To ensure the fire safety of visitors, structural features of buildings are essential; it is also crucial to take into account the psychophysiological characteristics of people in these buildings and their level of familiarity with the main structural elements of the building, willingness to find the fire sign and do their best to leave the building. The article describes promising fire safety systems for shopping centers, their advantages and disadvantages. A specialized system of fire extinguishing means consisting of interconnected software controls and structures and a gas fire extinguishing system and smoke curtains are described.

Design and architecture solutions for service robot waiter with a specialized system of payload stabilization

Application of robotic devices in subject domains, where monotonous routines have to be performed promptly and accurately, is a relevant problem, particularly in complicated epidemiological situations. In this paper the design of robot waiters is analyzed and a design is proposed for stabilization of payload during delivering. Common applied problems in the robotic service domain are associated with the need to use such robots on mostly even, flat surfaces or with the arrangement of special structures that simplify the movement of a robotic device along a given route. The proposed solution potentially provides for cheaper, simpler and more optimized application of the robotic device indoors, inside the restraunt due to the developed buffer mechanism and the system of gyroscopic stabilization of the trays, as well as the implemented control system based on the PID controller and the PWM generator, which ensures the smooth movement of the robot (from the starting point to the destination point). Based on the proposed solution, we get a fully functional robotic device that does not require additional investments in the reconstruction of the restaurant premises, completely replaces the waiter when delivering food and drinks to the client's table, and also attracts new customers due to its novelty and practicality.

A Specialized System for Arrhythmia Detection for Basic Research in Cardiology

The detection of cardiac arrhythmias has a long history in medicine, with current developments focusing on early detection using mobile devices. In basic research, however, the use cases and data differ greatly from the experimental setup. We developed a Python-based system to ease detection and analysis of arrhythmic sections in signals measured on extracted and stimulated cardiac myocytes. Multiple algorithms were integrated into the system, tested and evaluated. The best algorithm resulted in an F1-score of 0.97 and was primarily provided in the application.

A specialized system of pre-trip medical examinations using digital technologies

A system of pre-trip medical examinations using digital technologies has been developed. It includes a remote questionnaire, the results of which are analyzed and a reporting form is generated in text format and in the form of a graph. Portable, autonomous equipment and specialized software for use in remote medical examinations have been developed. Objective examination is carried out by the method of digital biomicroscopy of the eye, which allows you to analyze the parameters of pupillary reactions, the size of the palpebral fissure, the volume of eye movement, the frequency of blinking movements, the degree of redness of the eyeballs, the presence of nystagmus, the state of the surrounding eye tissues. Also in the hardware part of the complex there are sensors for detecting and measuring the level of concentration of ethyl alcohol vapors. Contact digital measurement of body temperature is performed with a sensor in contact with the forehead area during the examination. There is an additional program for the identification of the examined person by the iris of the eye. The use of digital methods in the medical examination system increases efficiency through the use of objective examination methods.

SALIVA CRYSTALLIZATION IN PATIENTS WITH COMBINED GASTROENTEROLOGICAL AND CARDIOVASCULAR PATHOLOGY

The purpose of this study was to estimate the specifity of saliva crystallization in comorbid pathology of digestive and cardiovascular systems. We collected saliva from 35 patients with coronary artery disease, 48 patients - with ulcer disease and 112 patients - with their combination. The features of self and initiated crystallization in the patients of all groups were evaluated using a specialized system of semi-quantitative parameters. It was stated that the crystallograms of comorbid patients are characterized by a clearly visible predominance of single-crystal elements over dendritic ones. The total density of structures in the facias is significantly reduced in comparison with both healthy people and patients of other groups.

Exploring source/goal asymmetries in spontaneous and caused motion expression in Yuhup

The aim of this paper is to explore the asymmetry in the expression of Goal and Source in spontaneous and caused motion events in Yuhup (Makú/Naduhup family). We examine the asymmetries at the semantic, morphological and syntactic levels. As this language does not have a specialized system of nominal morphology that distinguishes the ground constituents representing the Source or the Goal of a movement, an asymmetrical treatment of Source and Goal is not evident. However, verbal semantics make more distinctions for Goal than for Source, ground constituents are more frequent in the expression of motion to(wards) the Goal than the one (away) from the Source, and structural variety and complexity differentiate between Source and Goal expression. Also, the Goal bias is stronger in the descriptions of caused motion than in the ones of spontaneous motion. The data were collected using tools designed for elicitation of spontaneous and caused motion events.

The mental representation of sounds in speech sound disorders

The objective of this study is to investigate facets of the human phonological system in an attempt to elucidate the special nature of mental representations and operations underlying some of the errors in speech sound disorders (SSDs). After examining different theories on the mental representations of sounds and their organization in SSDs, we arrive at the conclusion that the existing elucidations on the phonological representations do not suffice to explain some distinctive facets of SSDs. Here, we endorse a hypothesis in favor of representationalism but offer an alternative conceptualization of the phonological representations (PR). We argue that the PR is to be understood in terms of a phonological base that holds information about a segment’s acoustic structure, and which interacts with other levels in the speech sound system in the mind so as to produce a certain sound. We also propose that the PR is connected to an interface module which mediates interactions between the PR and the articulatory system (AS) responsible for the physical manifestation of speech sounds in real time by way of the coordination of activities of speech organs in the vocal tract. We specifically consider different stages of operations within the interface, a specialized system within the cognitive system, which can explain patterns in the SSD data that have so far remained elusive. Positioned between the PR and the AS, the interface module is the heart of the current study. The presence of an interface module is necessitated by the fact that not all errors of SSDs are explainable in terms of structural, motor or even the symbolic misrepresentations at the level of PR. The interface acts as a mediating system mapping sound representations onto articulatory instructions for the actual production of sounds. The interface module can receive, process, and share the phonological inputs with other levels within the speech sound system. We believe an interface module such as ours holds the key to explaining at least certain speech disarticulations in SSDs.

A Specialized System-on-Chip Based Distance Protection for Distribution Grids with Inverter Interfaced Distributed Generation

Legacy protection schemes face new challenges as Inverter Interfaced Distribution Generation (IIDG) significantly changes the transient fault response of the distribution grid. The performance of the protection near the IIDG side is adversely affected by Low Voltage Ride Through (LVRT) and the negative sequence current suppression control characteristics of the IIDG. The operational characteristics of the protection are very different from those of the legacy protection schemes used in the distribution grid. Traditional overcurrent protection schemes cannot meet the requirements of selectivity and sensitivity. This paper analyses the influence of the IIDG on the protection schemes used in the distribution grid. Based on the positive sequence voltage polarization impedance criterion this study proposes a polarizing impedance criterion based on the fault component of positive sequence voltage, which can adaptively follow the fault resistance variation to satisfy the requirements of grid operation. The simulation results show that: (a) the proposed criterion is immune to the adverse effects of the transient characteristics of the IIDG; and, (b) it can adaptively follow the change of fault resistance making it suitable for application in short distribution lines. Using specialized System-on-Chip technology, a new distance protection device has been developed and tested on an industrial site. Simulation results and field tests showed that the new distance protection meets the requirements of the distribution grid with IIDGs.

Gas trapping of hot dust around main-sequence stars

ABSTRACT In 2006, Vega was discovered to display excess near-infrared (NIR) emission. Surveys now detect this phenomenon for one fifth of main-sequence stars, across various spectral types and ages. The excesses are interpreted as populations of small, hot dust grains very close to their stars, which must originate from comets or asteroids. However, the presence of such grains in copious amounts is mysterious, since they should rapidly sublimate or be blown out of the system. Here, we investigate a potential mechanism to generate excesses: dust migrating inwards under radiation forces sublimates near the star, releasing modest quantities of gas that then traps subsequent grains. This mechanism requires neither specialized system architectures nor high dust supply rates, and could operate across diverse stellar types and ages. The model naturally reproduces many features of inferred dust populations, in particular their location, preference for small grains, steep size distribution, and dust location scaling with stellar luminosity. For Sun-like stars, the mechanism can produce ${2.2 \; \mu {\rm m}}$ excesses that are an order of magnitude larger than those at ${8.5 \; \mu {\rm m}}$, as required by observations. However, for A-type stars the simulated NIR excesses were only twice those in the mid-infrared; grains would have to be 5–10 times smaller than those trapped in our model to be able to explain observed NIR excesses around A stars. Further progress with any hot dust explanation for A stars requires a means for grains to become very hot without either rapidly sublimating or being blown out of the system.