Small Number Of Examples Research Articles

Transferable and data efficient metamodeling of storm water system nodal depths using auto-regressive graph neural networks

Storm water systems (SWSs) are essential infrastructure providing multiple services including environmental protection and flood prevention. Typically, utility companies rely on computer simulators to properly design, operate, and manage SWSs. However, multiple applications in SWSs are highly time-consuming. Researchers have resorted to cheaper-to-run models, i.e. metamodels, as alternatives of computationally expensive models. With the recent surge in artificial intelligence applications, machine learning has become a key approach for metamodelling urban water networks. Specifically, deep learning methods, such as feed-forward neural networks, have gained importance in this context. However, these methods require generating a sufficiently large database of examples and training their internal parameters. Both processes defeat the purpose of using a metamodel, i.e., saving time. To overcome this issue, this research focuses on the application of inductive biases and transfer learning for creating SWS metamodels which require less data and retain high performance when used elsewhere. In particular, this study proposes an auto-regressive graph neural network metamodel of the Storm Water Management Model (SWMM) from the Environmental Protection Agency (EPA) for estimating hydraulic heads. The results indicate that the proposed metamodel requires a smaller number of examples to reach high accuracy and speed-up, in comparison to fully connected neural networks. Furthermore, the metamodel shows transferability as it can be used to predict hydraulic heads with high accuracy on unseen parts of the network. This work presents a novel approach that benefits both urban drainage practitioners and water network modeling researchers. The proposed metamodel can help practitioners on the planning, operation, and maintenance of their systems by offering an efficient metamodel of SWMM for computationally intensive tasks like optimization and Monte Carlo analyses. Researchers can leverage the current metamodel’s structure for developing new surrogate model architectures tailored to their specific needs or start paving the way for more general foundation metamodels of urban drainage systems.

RetClean: Retrieval-Based Data Cleaning Using LLMs and Data Lakes

Large language models (LLMs) have shown great potential in data cleaning, which is a fundamental task in all modern applications. In this demo proposal, we demonstrate that indeed LLMs can assist in data cleaning, e.g., filling in missing values in a data table, through different approaches. For example, cloud-based non-private LLMs, e.g., OpenAI GPT family or Google Gemini, can assist in cleaning non-private datasets that encompass world-knowledge information (Scenario 1). However, such LLMs may struggle with datasets that they have never encountered before, e.g., local enterprise data, or when the user requires an explanation of the source of the suggested clean values. In that case, retrieval-based methods using RAG (Retrieval Augmented Generation) that complements the LLM power with a user-provided data source, e.g., a data lake, are a must. The data lake is indexed, and each time a new request comes, we retrieve the top- k relevant tuples to the user's query tuple to be cleaned and leverage LLM inference power to infer the correct value (Scenario 2). Nevertheless, even in Scenario 2, sharing enterprise data with public LLMs (an externally hosted model) might not be feasible for privacy reasons. In this scenario, we showcase the practicality of locally hosted small LLMs in the cleaning process, especially after fine-tuning them on a small number of examples (Scenario 3). Our proposed system, RetClean , seamlessly supports all three scenarios and provides a user-friendly GUI that enables the VLDB audience to explore and experiment with different LLMs and investigate their trade-offs.

State Machines for Large Scale Computer Software and Systems

The behavior and architecture of large scale discrete state systems found in computer software and hardware can be specified and analyzed using a particular class of primitive recursive functions. This article begins with an illustration of the utility of the method via a number of small examples and then via longer specification and verification of the “Paxos” distributed consensus algorithm [ 26 ]. The “sequence maps” are then shown to provide an alternative representation of deterministic state machines and products of state machines. Distributed and composite systems, parallel and concurrent computation, and real-time behavior can all be specified naturally with these methods—which require neither extensions to the classical state machine model nor any axiomatic methods or other techniques from formal logic or other foundational methods. Compared to state diagrams or tables or the standard set-tuple-transition-maps, sequence maps are more concise and better suited to describing the behavior and compositional architecture of computer systems. Staying strictly within the boundaries of classical deterministic state machines anchors the methods to the algebraic structures of automata and makes the specifications faithful to engineering practice.

Exploring generative AI assisted feedback writing for students’ written responses to a physics conceptual question with prompt engineering and few-shot learning

Instructor’s feedback plays a critical role in students’ development of conceptual understanding and reasoning skills. However, grading student written responses and providing personalized feedback can take a substantial amount of time, especially in large enrollment courses. In this study, we explore using GPT-3.5 to write feedback on students’ written responses to conceptual questions with prompt engineering and few-shot learning techniques. In stage I, we used a small portion (n=20) of the student responses on one conceptual question to iteratively train GPT to generate feedback. Four of the responses paired with human-written feedback were included in the prompt as examples for GPT. We tasked GPT to generate feedback for another 16 responses and refined the prompt through several iterations. In stage II, we gave four student researchers (one graduate and three undergraduate researchers) the 16 responses as well as two versions of feedback, one written by the authors and the other by GPT. Students were asked to rate the correctness and usefulness of each feedback and to indicate which one was generated by GPT. The results showed that students tended to rate the feedback by human and GPT equally on correctness, but they all rated the feedback by GPT as more useful. Additionally, the success rates of identifying GPT’s feedback were low, ranging from 0.1 to 0.6. In stage III, we tasked GPT to generate feedback for the rest of the students’ responses (n=65). The feedback messages were rated by four instructors based on the extent of modification needed if they were to give the feedback to students. All four instructors rated approximately 70% (ranging from 68% to 78%) of the feedback statements needing only minor or no modification. This study demonstrated the feasibility of using generative artificial intelligence (AI) as an assistant to generate feedback for student written responses with only a relatively small number of examples in the prompt. An AI assistant can be one of the solutions to substantially reduce time spent on grading student written responses. Published by the American Physical Society 2024

Perspectives on the application of large language models in healthcare

Large language models have become a new significant breakthrough in the field of artificial intelligence. They are changing approaches to machine learning from models that solve narrow problems and require large amounts of data with the known answers for training, to generative models that are fine tunable to solve specific problems using a small number of examples with the known answers or even none at all. Medicine is one of the areas in which the use of large language models can become extremely useful. The review presents data on the latest achievements in the use of large language models for medical tasks, prospects for using these models as the basis for the digital assistants for doctors and patients, as well as existing regulatory and ethical barriers to the development of this breakthrough technology for addressing healthcare challenges.

Generalisation through Negation and Predicate Invention

The ability to generalise from a small number of examples is a fundamental challenge in machine learning. To tackle this challenge, we introduce an inductive logic programming (ILP) approach that combines negation and predicate invention. Combining these two features allows an ILP system to generalise better by learning rules with universally quantified body-only variables. We implement our idea in NOPI, which can learn normal logic programs with predicate invention, including Datalog programs with stratified negation. Our experimental results on multiple domains show that our approach can improve predictive accuracies and learning times.

Increased Usage of Alt Text Is Required Across Ontario Public Library Social Media Feeds to Increase the Accessibility of Content

A Review of: Hill, H., & Oswald, K. (2023). “May be a picture of a dog and a book”: The inaccessibility of public libraries’ social media feeds. Partnership, 18(1), 1–14. https://doi.org/10.21083/partnership.v18i1.7008 Objective – The research project sought to explore how accessible the social media feeds of Ontario public libraries are, particularly the use of alt text for images, by assessing the usage of alt text and by making recommendations for appropriate use within social media posts. Design – Collection of social media posts and computer-assisted textual analysis of visual media content. Setting – 76 public libraries and 9 public library systems in Ontario, Canada. Subjects – Approximately 900 Ontario public library social media posts from Facebook, Twitter, and Instagram. Methods – A random number generator sampling of 30 libraries per platform from the relevant social media accounts from a spreadsheet created using Ontario Public Library Statistics (OPLS) data of social media usage from the included libraries was initially created capturing 76 individual libraries. Then the researchers performed targeted sampling of posts from the nine library systems serving over 250,000 residents each. Researchers identified the 10 most recent posts from each included platform feed, and then undertook textual analysis for the presence of alt text with each post using two Mozilla Firefox browser extensions that determine the presence of alt text. Main Results – Of the 76 unique libraries chosen by the random sampling and the nine library systems that serve populations over 250,000, only two regularly used alt text and five had at least one instance of alt text. Only Toronto Public Library regularly included alt text across each of the three social media platforms analyzed by the study. The study also initially aimed to assess the quality of alt text used by public libraries in social media posts. However, due to the lack of alt text use across the sample, this was not possible at the scale initially aimed for, although a small number of examples are analyzed in the findings. Conclusion – The initial goal of analyzing the alt text to make recommendations for improved usage could not be realized due to the surprising lack of inclusion of any alt text across the sampled posts. This lack of any alt text can prevent some disabled users from engaging with content and information, leading to an inequitable experience. Public libraries should consider how accessible their engagement with users is and seek to improve the accessibility of social media posts.

Investigating the ability of deep learning-based structure prediction to extrapolate and/or enrich the set of antibody CDR canonical forms.

Deep learning models have been shown to accurately predict protein structure from sequence, allowing researchers to explore protein space from the structural viewpoint. In this paper we explore whether "novel" features, such as distinct loop conformations can arise from these predictions despite not being present in the training data. Here we have used ABodyBuilder2, a deep learning antibody structure predictor, to predict the structures of ~1.5M paired antibody sequences. We examined the predicted structures of the canonical CDR loops and found that most of these predictions fall into the already described CDR canonical form structural space. We also found a small number of "new" canonical clusters composed of heterogeneous sequences united by a common sequence motif and loop conformation. Analysis of these novel clusters showed their origins to be either shapes seen in the training data at very low frequency or shapes seen at high frequency but at a shorter sequence length. To evaluate explicitly the ability of ABodyBuilder2 to extrapolate, we retrained several models whilst withholding all antibody structures of a specific CDR loop length or canonical form. These "starved" models showed evidence of generalisation across CDRs of different lengths, but they did not extrapolate to loop conformations which were highly distinct from those present in the training data. However, the models were able to accurately predict a canonical form even if only a very small number of examples of that shape were in the training data. Our results suggest that deep learning protein structure prediction methods are unable to make completely out-of-domain predictions for CDR loops. However, in our analysis we also found that even minimal amounts of data of a structural shape allow the method to recover its original predictive abilities. We have made the ~1.5 M predicted structures used in this study available to download at https://doi.org/10.5281/zenodo.10280181.

Few-shot code translation via task-adapted prompt learning

Pre-trained models such as CodeT5 and TransCoder have achieved impressive progress in software engineering. However, fine-tuning PLMs for code translation is confronted with significant challenges owing to the scarce availability of parallel code. Large language models such as ChatGPT have exhibited considerable promise in few-shot learning where only a small number of demonstration examples are given to the LLM. Yet they have not been specifically optimized for domain-specific tasks, and their use often entails significant manual effort in manually curating prompts. In this paper, we propose FSCTrans, a novel parameter-efficient tuning approach for code translation when furnished with only a few demonstration examples. (1) to efficiently reuse prior knowledge during pre-training, FSCTrans employs task-adapted prompt tuning, which freezes the pre-trained CodeT5 while merely updating parameters in a small prompt module; (2) to enable parameter efficient tuning on only a small number of examples, FSCTrans bridges pre-training to the translation task through a new pre-training objective of code-to-code generation. We evaluate FSCTrans on Java↔Python and Java↔C# datasets from both real-world projects and online judge problems. The evaluation results show that FSCTrans is remarkably effective in few-shot code translation: on average, it improves CodeT5 by 54.61% and 31.59% in terms of BLEU-4 and CodeBLEU; notably, FSCTrans demonstrates 14.42% and 18.36% superior performance in Java → C# translations in terms of BLEU-4 and CodeBLEU compared to ChatGPT.

The role of lifelong machine learning in bridging the gap between human and machine learning: A scientometric analysis

AbstractDue to advancements in data collection, storage, and processing techniques, machine learning has become a thriving and dominant paradigm. However, one of its main shortcomings is that the classical machine learning paradigm acts in isolation without utilizing the knowledge gained through learning from related tasks in the past. To circumvent this, the concept of Lifelong Machine Learning (LML) has been proposed, with the goal of mimicking how humans learn and acquire cognition. Human learning research has revealed that the brain connects previously learned information while learning new information from a single or small number of examples. Similarly, an LML system continually learns by storing and applying acquired information. Starting with an analysis of how the human brain learns, this paper shows that the LML framework shares a functional structure with the brain when it comes to solving new problems using previously learned information. It also provides a description of the LML framework, emphasizing its similarities to human brain learning. It also provides citation graph generation and scientometric analysis algorithms for the LML literatures, including information about the datasets and evaluation metrics that have been used in the empirical evaluation of LML systems. Finally, it presents outstanding issues and possible future research directions in the field of LML.This article is categorized under: Technologies > Machine Learning

Prevention of emergency situations in complex technical systems using a neuromorphic approach

The paper proposes a scheme of emergency prevention system based on neuromorphic approach. The system includes a prediction unit that implements a mathematical model of the cerebellum predictive functions, and an alarm unit that implements the pain sensation model, proposed by the authors earlier. As a basic element of the proposed system the Compartmental Spiking Neuron Model (CSNM) was used, capable of learning from a small number of examples. The use of neuromorphic approach allows to overcome the limitations associated with the formalizing complexity of the systems being diagnosed and the low availability of data for modeling the processes occurring in them. The overcoming these limitations is possible due to the possibility of learning from a small number of examples and the absence of the need to model the system being diagnosed itself. The paper also presents the results of testing of the proposed scheme, which was carried out on a computer model using synthetic data. The results of the testing showed the fundamental applicability of the proposed scheme in neuromorphic control systems.

Analog RF Circuit Sizing by a Cascade of Shallow Neural Networks

A deep neural network architecture for the automatic sizing of analog circuit components is proposed, with a focus on radio frequency (RF) applications in the 2 to 5 GHz region. It addresses the challenges of the typically small number of examples for network training and the existence of multiple solutions, of which impractical values for integrated circuit implementation. We address these issues by restricting the learning to one component size at a time, thanks to a cascade of dedicated shallow neural networks (SNN), where each network constrains the prediction of the next ones. Moreover, the SNNs are individually tuned by a genetic algorithm for the prediction order and accuracy. This reduction of the solution space at each step allows the use of small training sets, and the introduced constraints between SNNs handle component interdependencies. The method is successfully validated on three different types of RF microcircuits: a low-noise amplifier (LNA), a voltage-controlled oscillator (VCO), and a mixer, using 180 nm and 130 nm CMOS implementations. All the predictions were within 5 % of the true values, both at the component and performance levels, and all the responses were obtained in less than 5 s, after 4 to 47 min. training on a regular PC station. The obtained results show that the proposed method is fast and applicable to arbitrary analog circuit topologies, with no need to retrain the developed neural network for each new set of desired circuit performances.

Focus on age at death in field epidemiology.

Age at death (AD) is an old metric recently re-evaluated for the study of longevity and mainly used in demography. Developed experience using AD in field epidemiology is summarized with cohorts followed-up for variable periods of time, frequently until extinction or close to extinction, a must to correctly adopt this metric. For practical purposes, a small number of examples is reported condensing previously published results to highlight various aspects of the problem. AD became the alternative of overall death rates when comparing cohorts reaching extinction or near extinction. AD was useful to characterize different causes of death in order to describe their natural history and possible etiology. With the use of multiple linear regression, a large number of possible determinants of AD were identified and some combinations of them resulted in large estimated differences in AD of 10years or more across individuals. AD is a powerful tool to study population samples followed-up until extinction or near extinction. It allows to compare the life-long experience of different populations, to compare the role of different causes of death and to study the determinants of AD that are conditioning longevity.

The effect of visitors on the behaviour of zoo-housed primates: A test of four hypotheses

The effects of human visitors on the behaviour and welfare of zoo-housed primates have been extensively studied, and we know more about these effects in primates than in any other taxon. Nevertheless, it has proved difficult to detect an overall pattern in the responses of primates to visitors, as the presence of people can be stressful to the animals, but in some circumstances can be enriching, and in other circumstances the animals appear to be largely unaffected. Several potential causal factors have been suggested, namely that variability is due to species-specific differences, differences in visitor characteristics and behaviour, differences in housing and husbandry, or differences in individual characteristics of the animals, and these can be treated as four hypotheses to account for this variability in response. Here we test these four hypotheses using published data. A significant association was found between animal response and ecological category, such that small arboreal primates were most likely to be affected negatively. Visitor characteristics showed a significant association with animal response with noisy visitors having the most stressful effect. There was no significant effect of enclosure size or group size, though there was a significant association with enclosure type, such that animals in walk-through and semi-free range enclosures were more likely to show positive responses, though the small number of examples of these suggests treating this result with caution. Finally, individual differences were tested with gorillas, where there were sufficient data to look at individuals rather than animal groups. There was no significant effect of age or sex, but regression using all of the potential causal variables produced a significant model in which no individual predictor was significant, suggesting that it is the combination of predictors which influences the form of the response.

Prospects of cooperation of the Russian Federation with friendly countries in the framework of investment activities.

The object of the study is investment activity in the Russian Federation. The subject of the study is the implementation of investment activities in relation to friendly foreign states. The purpose of the work is to assess the possible prospects and risks of the orientation of the economy of the Russian Federation on cooperation with friendly countries in the field of investment. Tasks of the work: • To assess the regulatory framework for the orientation of the economy of the Russian Federation to friendly countries; • Analyze economic indicators in relation to friendly countries; • To assess the prospects for cooperation with unfriendly countries in the medium term. The main method of research is analysis. The author analyzes the regulatory framework in the sphere of orientation of the economy of the Russian Federation to strengthen partnership with friendly countries, provides analysis of various indicators of friendly countries and companies belonging to the jurisdiction of friendly countries. Based on the analysis, the author identifies the problem points, relying on the methods of logic. Today, the economy of the Russian Federation is reorienting towards building relations with friendly states. Such a decision is an atypical measure of economic regulation, and there are a small number of examples of similar examples of actions by state bodies. The decision causes an active discussion in the Russian society, which confirms the relevance of this study. The key problem points directly related to the restriction of access to investments in companies of unfriendly foreign states and current investment opportunities in companies of friendly states are identified. The author puts forward several recommendations that can reduce the level of uncertainty in the financial market and reduce risks in the field of investments.

Trust, agreements, and occasional breakdowns: Veterinarians' perspectives on farmer-veterinarian relationships and use of antimicrobials for Swedish dairy cattle.

Studies have shown that farmer-veterinarian relationships influence antimicrobial use (AMU) in livestock, though how they do so is unclear. On the one hand, research shows that well-established veterinarian-farmer relationships are positive for implementation of antibiotic stewardship and restrictive AMU. On the other hand, studies also show that farmer demands can increase antimicrobial prescribing and that prescribing antimicrobials can strengthen the veterinarian's relationship with farmer clients. In the present study, we focus on veterinarians' perspectives on the relationships between dairy cattle farmers and veterinarians in Sweden and explore what characterizes these relationships when restrictive AMU is described as unproblematic and when AMU becomes a matter of tension or conflict. The study draws on semistructured interviews with 21 veterinarians working with livestock in Sweden. Interviews were analyzed thematically. The study shows that from the perspectives of veterinarians, well-established veterinarian-farmer relationships generally facilitate restrictive AMU in 3 slightly different but related ways: (1) they create trust in the veterinarian and their prescribing decisions; (2) they create shared understanding concerning when antimicrobials are needed and not needed; and (3) they facilitate constructive discussions between veterinarians and farmers on AMU. To make the farmer feel listened to and to come to an agreement on AMU was described as central for the veterinarians. However, the veterinarians described agreements on restrictive AMU as sometimes requiring strategic work, such as discussions to motivate the farmer and leave the door open for antimicrobials later if needed. Such work takes time and energy and is easier within well-established relationships according to the veterinarians. We also identified examples where veterinarians explained that they occasionally make compromises with farmers concerning antimicrobials-compromises that, according to the veterinarians, facilitate the relationship with the farmer, which in turn facilities restrictive AMU in the longer term. The examples in our interviews where antimicrobials became a matter of tension and even conflict between veterinarians and farmers could, with a few exceptions, be traced to absence of well-established relationships. However, some veterinarians also described AMU as a matter of tension within well-established relationships, especially with older farmers who do not trust new treatment strategies. We also found a small number of examples where disagreements on antimicrobials made a relationship with specific farmers impossible. Thus, even though the interviewed veterinarians generally meant that their restrictive antimicrobial prescribing did not threaten the relationship with the farmer, our study also, to some extent, confirms research showing that restrictive AMU can harm the relationship with some clients who simply choose another, less restrictive, veterinarian. In summary, our study supports that decisions on AMU cannot be reduced to the individual prescriber's behavior, nor to a strict medical judgment. Antimicrobial use for dairy cattle needs to be understood as taking form in relationships in which both veterinarians and farmers are active parts.

Exploiting Domain Knowledge to Address Class Imbalance in Meteorological Data Mining

We deal with the problem of class imbalance in data mining and machine learning classification algorithms. This is the case where some of the class labels are represented by a small number of examples in the training dataset compared to the rest of the class labels. Usually, those minority class labels are the most important ones, implying that classifiers should primarily perform well on predicting those labels. This is a well-studied problem and various strategies that use sampling methods are used to balance the representation of the labels in the training dataset and improve classifier performance. We explore whether expert knowledge in the field of Meteorology can enhance the quality of the training dataset when treated by pre-processing sampling strategies. We propose four new sampling strategies based on our expertise on the data domain and we compare their effectiveness against the established sampling strategies used in the literature. It turns out that our sampling strategies, which take advantage of expert knowledge from the data domain, achieve class balancing that improves the performance of most classifiers.

Art in Surgery: A Review of Art-based Medical Humanities Curricula in Surgical Residency

Numerous programs integrate arts and humanities methods to advance medical education competencies. Despite the highly visual and technical nature of the field of surgery, the current state of art utilization in surgical training is unclear. The purpose of this review is to gain a comprehensive understanding of how art has been utilized in surgical training, to investigate the purpose of such interventions, and to assess how art interventions may benefit surgeons. A systematic literature review using PRISMA methodology was conducted to identify articles published prior to February 2022 that investigated or described using art in surgical resident training. Qualitative themes were developed upon full review of the literature and categorized based on fundamental aspects of surgical education. The data was summarized by a narrative approach. Six hundred seventy-four unique articles were initially identified, thirteen of which met inclusion criteria. Twelve studies employed drawing or sculpture in surgical residency training; one discussed art observation to foster mindfulness, teambuilding, and empathy. Eight articles utilized art as an evaluation tool, 2 for didactic and archival purposes, one employed exercises in art analysis to improve empathy and physician wellbeing, and 2 described courses in which art making was treated as a foundational skill. No articles discussed use of art for honing diagnostic skills, observation, or patient communication - competencies that have been addressed in other fields. This review highlights the small number of examples in the medical literature about visual arts in surgical training. The existing art-based surgical humanities studies identify opportunities for curricular innovation within surgical training.

Baboon perspectives on the ecology and behavior of early human ancestors

For more than 70 y researchers have looked to baboons (monkeys of the genus Papio) as a source of hypotheses about the ecology and behavior of early hominins (early human ancestors and their close relatives). This approach has undergone a resurgence in the last decade as a result of rapidly increasing knowledge from experimental and field studies of baboons and from archeological and paleontological studies of hominins. The result is a rich array of analogies, scenarios, and other stimuli to thought about the ecology and behavior of early hominins. The main intent here is to illustrate baboon perspectives on early hominins, with emphasis on recent developments. This begins with a discussion of baboons and hominins as we know them currently and explains the reasons for drawing comparisons between them. These include occupation of diverse environments, combination of arboreal and terrestrial capabilities, relatively large body size, and sexual dimorphism. The remainder of the paper illustrates the main points with a small number of examples drawn from diverse areas of interest: diet (grasses and fish), danger (leopards and crocodiles), social organization (troops and multilevel societies), social relationships (male–male, male–female, female–female), communication (possible foundations of language), cognition (use of social information, comparison of self to others), and bipedalism (a speculative developmental hypothesis about the neurological basis). The conclusion is optimistic about the future of baboon perspectives on early hominins.

MULTILINGUAL TEXT CLASSIFIER USING PRE-TRAINED UNIVERSAL SENTENCE ENCODER MODEL

Context. Online platforms and environments continue to generate ever-increasing content. The task of automating the moderation of user-generated content continues to be relevant. Of particular note are cases in which, for one reason or another, there is a very small amount of data to teach the classifier. To achieve results under such conditions, it is important to involve the classifier pre-trained models, which were trained on a large amount of data from a wide range. This paper deals with the use of the pre-trained multilingual Universal Sentence Encoder (USE) model as a component of the developed classifier and the affect of hyperparameters on the classification accuracy when learning on a small data amount (~ 0.05% of the dataset).
 Objective. The goal of this paper is the investigation of the pre-trained multilingual model and optimal hyperparameters influence for learning the text data classifier on the classification result.
 Method. To solve this problem, a relatively new approach to few-shot learning has recently been used – learning with a relatively small number of examples. Since text data is still the dominant way of transmitting information, the study of the possibilities of constructing a classifier of text data when learning from a small number of examples (~ 0.002–0.05% of the data set) is an actual problem.
 Results. It is shown that even with a small number of examples for learning (36 per class) due to the use of USE and optimal configuration in learning can achieve high accuracy of classification on English and Russian data, which is extremely important when it is impossible to collect your own large data set. The influence of the approach using USE and a set of different configurations of hyperparameters on the result of the text data classifier on the example of English and Russian data sets is evaluated.
 Conclusions. During the experiments, a significant degree of relevance of the correct selection of hyperparameters is shown. In particular, this paper considered the batch size, optimizer, number of learning epochs and the percentage of data from the set taken to train the classifier. In the process of experimentation, the optimal configuration of hyperparameters was selected, according to which 86.46% accuracy of classification on the Russian-language data set and 91.13% on the English-language data, respectively, can be achieved in ten seconds of training (training time can be significantly affected by technical means used).