Search

The insertion of Machine Learning (ML) in everyday life demonstrates the importance of popularizing an understanding of ML already in school. Accompanying this trend arises the need to assess the students' learning. Yet, so far, few assessments have been proposed, most lacking an evaluation. Therefore, we evaluate the reliability and validity of an automated assessment of the students' learning of an image classification model created as a learning outcome of the "ML for All!" course. Results based on data collected from 240 students indicate that the assessment can be considered reliable (coefficient Omega = 0.834/Cronbach's alpha [alpha] = 0.83). We also identified moderate to strong convergent and discriminant validity based on the polychoric correlation matrix. Factor analyses indicate two underlying factors "Data Management and Model Training" and "Performance Interpretation", completing each other. These results can guide the improvement of assessments, as well as the decision on the application of this model in order to support ML education as part of a comprehensive assessment.

Although Machine Learning (ML) is found practically everywhere, few understand the technology behind it. This presents new challenges to extend computing education by including ML concepts in order to help students to understand its potential and limits and empowering them to become creators of intelligent solutions. Therefore, we developed an introductory course to teach basic ML concepts, such as fundamentals of neural networks as well as limitations and ethical concerns in alignment with the K-12 Guidelines for Artificial Intelligence. It also teaches the application of these concepts, by guiding the students to develop a first classification model of recycling trash images using Google Teachable Machine. In order to promote ML education, the interactive course is available online in Brazilian Portuguese to be used as an extracurricular course or in an interdisciplinary way as part of science classes covering recycling topics. The course was applied and evaluated through a series of exploratory case studies with a total of 108 middle and high school students. The results of the evaluation show that middle and high school students were able to understand ML concepts and develop an image classification model. No substantial difference with regard to the educational stage, gender and instructional mode was observed. The course was also perceived as an enjoyable learning experience motivating students to learn more about ML.

Although Machine Learning (ML) has already become part of our daily lives, few are familiar with this technology. Thus, in order to help students to understand ML, its potential, and limitations and to empower them to become creators of intelligent solutions, diverse courses for teaching ML in K-12 have emerged. Yet, a question less considered is how to assess the learning of ML. Therefore, we performed a systematic mapping identifying 27 instructional units, which also present a quantitative assessment of the students' learning. The simplest assessments range from quizzes to performance-based assessments assessing the learning of basic ML concepts, approaches, and in some cases ethical issues and the impact of ML on lower cognitive levels. Feedback is mostly limited to the indication of the correctness of the answers and only a few assessments are automated. These results indicate a need for more rigorous and comprehensive research in this area.

The insertion of Machine Learning (ML) in everyday life demonstrates the importance of popularizing an understanding of ML already in school. Accompanying this trend arises the need to assess the students’ learning. Yet, so far, few assessments have been proposed, most lacking an evaluation. Therefore, we evaluate the reliability and validity of an automated assessment of the students’ learning of an image classification model created as a learning outcome of the “ML for All!” course. Results based on data collected from 240 students indicate that the assessment can be considered reliable (coefficient Omega = 0.834/Cronbach's alpha α=0.83). We also identified moderate to strong convergent and discriminant validity based on the polychoric correlation matrix. Factor analyses indicate two underlying factors “Data Management and Model Training” and “Performance Interpretation”, completing each other. These results can guide the improvement of assessments, as well as the decision on the application of this model in order to support ML education as part of a comprehensive assessment.

Although Machine Learning (ML) is used already in our daily lives, few are familiar with the technology. This poses new challenges for students to understand ML, its potential, and limitations as well as to empower them to become creators of intelligent solutions. To effectively guide the learning of ML, this article proposes a scoring rubric for the performance-based assessment of the learning of concepts and practices regarding image classification with artificial neural networks in K-12. The assessment is based on the examination of student-created artifacts as a part of open-ended applications on the use stage of the Use-Modify-Create cycle. An initial evaluation of the scoring rubric through an expert panel demonstrates its internal consistency as well as its correctness and relevance. Providing a first step for the assessment of concepts on image recognition, the results may support the progress of learning ML by providing feedback to students and teachers.

The poor representation of the macrophysical properties of shallow oceanic cumuli in climate models contributes to the large uncertainty in cloud feedback. These properties are also difficult to measure because it requires high-resolution satellite imagery that is seldomly collected over ocean. Here, we examine cumulus cloud macrophysical properties, their size, shape, and spatial distributions, over the tropical western Pacific using 170 15 m resolution scenes from Terra's Advanced Spaceborne Thermal Emission and Reflection Radiometer (ASTER) collected during the 2019 Cloud, Aerosol and Monsoon Processes Philippines Experiment (CAMP2Ex) mission. The average cloud fraction (CF) was 0.12, half of which was contributed by clouds less than 1.6 km in area-equivalent diameter. This compared well to Terra's Multi-angle Imaging SpectroRadiometer (MISR) resolution-corrected CF of 0.14 but less than the 0.19 measured by Terra's Moderate Resolution Imaging Spectroradiometer (MODIS). The cloud size distribution exhibited a power law form with an exponent of 2.93 and an area–perimeter power law with a dimension of 1.25. ASTER, MISR, and CAMP2Ex aircraft lidar showed excellent agreement in the cloud top height (CTH) distribution peak altitude of ∼ 750 m. We examined cumulus properties in relation to meteorological variables and found that the variation in mean CTH is controlled most by the total column water vapor, lower-tropospheric stability (LTS), and estimated inversion strength (EIS). The variation in CF is most controlled by surface wind speed and near-cloud relative humidity instead of LTS/EIS, suggesting the need to improve low-cloud parameterizations in climate models that use LTS/EIS based on stratocumulus studies.

Abstract The oxidative potential (OP) of particulate matter (PM) is a major driver of PM-associated health effects. In India, the emission sources defining PM-OP, and their local/regional nature, are yet to be established. Here, to address this gap we determine the geographical origin, sources of PM, and its OP at five Indo-Gangetic Plain sites inside and outside Delhi. Our findings reveal that although uniformly high PM concentrations are recorded across the entire region, local emission sources and formation processes dominate PM pollution. Specifically, ammonium chloride, and organic aerosols (OA) from traffic exhaust, residential heating, and oxidation of unsaturated vapors from fossil fuels are the dominant PM sources inside Delhi. Ammonium sulfate and nitrate, and secondary OA from biomass burning vapors, are produced outside Delhi. Nevertheless, PM-OP is overwhelmingly driven by OA from incomplete combustion of biomass and fossil fuels, including traffic. These findings suggest that addressing local inefficient combustion processes can effectively mitigate PM health exposure in northern India.

ABSTRACT The objectives of the present study were to determine anatomical characteristics, mechanical properties and natural durability to two fungi types [Gloeophyllum trabeum (Persoon ex Fries) Murrill and Trametes versicolor (Linnaeus ex Fries)Pilat] for cypress wood (Cupressus spp.). The wood has straight grain, brown to white color, medium texture, and moderate luster. It also showed medium to low density, low shrinkage, and moderate dimensional stability. Lignin content was relatively high, whereas extractive content was low. Cupressus spp. showed static bending properties of 4.1 GPa for modulus of elasticity; 91.0 MPa for modulus of rupture; 1.8 MPa for internal bond; and 6357 and 4039 N for end and side Janka hardness, respectively. All mechanical properties, in air-dry condition, were similar to those described to cypresses and some Amazonian hardwood species. Cypress wood was ranked as highly resistant to the brown-rot fungus Gloeophyllum trabeum and to the white-rot fungus Trametes versicolor, making it suitable for interior and exterior uses and high humidity environment.

ABSTRACT Artificial insemination success in swine is mainly associated with semen dose quality. Thus, this study compared quality control parameters in 11 Brazilian boar studs after applying audit services for 24 months (1,650 boars). An extensive checklist was applied in each audit, registering ‘compliance’ or ‘noncompliance’ for 75 items. Semen doses produced were analyzed as regards volume and sperm concentration, and microbiological analyses were conducted for semen and water samples collected at distinct production stages. On average, boar studs produced 112.9 semen doses per boar per month, and the odds of raw semen contamination increased when boars were inadequately housed and doses were collected under increased temperatures, with no anti-slip rubber mat or after a poor prepuce cleaning (p < 0.05). Collection from boars with locomotor problems and no regular change of reverse osmosis filters increased the contamination odds in semen doses produced and stored at the stud (p < 0.05). As regards the water submitted to the osmosis reverse process, contamination odds increased as a result of deficient cleaning and disinfection of the purification equipment (p < 0.05). Risk factors for reduced sperm motility (< 70 %) were: no anti-slip rubber mat for semen collection, no cleaning program for automatic feeding system (drops) and bins, and inadequate intervals between semen collections (≤ 2 days or > 7 days; p < 0.05). Two boar studs had the best results for compliance with the checklist items. Constant monitoring, appropriate hygiene of facilities and equipment, and periodical staff training are highlighted as non-negotiable points for boar semen dose quality.

Trusted Research Environments (TREs) enable the analysis of sensitive data under strict security assertions that protect the data with technical, organizational, and legal measures from (accidentally) being leaked outside the facility. While many TREs exist in Europe, little information is available publicly on the architecture and descriptions of their building blocks and their slight technical variations. To highlight on these problems, an overview of the existing, publicly described TREs and a bibliography linking to the system description are provided. Their technical characteristics, especially in commonalities and variations, are analysed, and insight is provided into their data type characteristics and availability. The literature study shows that 47 TREs worldwide provide access to sensitive data, of which two-thirds provide data predominantly via secure remote access. Statistical offices (SOs) make the majority of sensitive data records included in this study available.

We analyzed long-term measurements of organic carbon, elemental carbon, and source-specific organic tracers from 2017 to 2020 to constrain carbonaceous aerosol sources in the rapidly changing Arctic. Additionally, we used absorption photometer (Aethalometer) measurements to constrain equivalent black carbon (eBC) from biomass burning and fossil fuel combustion, using positive matrix factorization (PMF). Our analysis shows that organic tracers are essential for understanding Arctic carbonaceous aerosol sources. Throughout 2017 to 2020, levoglucosan exhibited bimodal seasonality, reflecting emissions from residential wood combustion (RWC) in the heating season (November to May) and from wildfires (WFs) in the non-heating season (June to October), demonstrating a pronounced interannual variability in the influence of WF. Biogenic secondary organic aerosol (BSOA) species (2-methyltetrols) from isoprene oxidation was only present in the non-heating season, peaking in July to August. Warm air masses from Siberia led to a substantial increase in 2-methyltetrols in 2019 and 2020 compared to 2017 to 2018. This highlights the need to investigate the contribution of local sources vs. long-range atmospheric transport (LRT), considering the temperature sensitivity of biogenic volatile organic compound emissions from Arctic vegetation. Tracers of primary biological aerosol particles (PBAPs), including various sugars and sugar alcohols, showed elevated levels in the non-heating season, although with different seasonal trends, whereas cellulose had no apparent seasonality. Most PBAP tracers and 2-methyltetrols peaked during influence of WF emissions, highlighting the importance of measuring a range of source-specific tracers to understand sources and dynamics of carbonaceous aerosol. The seasonality of carbonaceous aerosol was strongly influenced by LRT episodes, as background levels are extremely low. In the non-heating season, the organic aerosol peak was as influenced by LRT, as was elemental carbon during the Arctic haze period. Source apportionment of carbonaceous aerosol by Latin hypercube sampling showed mixed contributions from RWC (46 %), fossil fuel (FF) sources (27 %), and BSOA (25 %) in the heating season. In contrast, the non-heating season was dominated by BSOA (56 %), with lower contributions from WF (26 %) and FF sources (15 %). Source apportionment of eBC by PMF showed that FF combustion dominated eBC (70±2.7 %), whereas RWC (22±2.7 %) was more abundant than WF (8.0±2.9 %). Modeled BC concentrations from FLEXPART (FLEXible PARTicle dispersion model) attributed an almost equal share to FF sources (51±3.1 %) and to biomass burning. Both FLEXPART and the PMF analysis concluded that RWC is a more important source of (e)BC than WF. However, with a modeled RWC contribution of 30±4.1 % and WF of 19±2.8 %, FLEXPART suggests relatively higher contributions to eBC from these sources. Notably, the BB fraction of EC was twice as high as that of eBC, reflecting methodological differences between source apportionment by LHS and PMF. However, important conclusions drawn are unaffected, as both methods indicate the presence of RWC- and WF-sourced BC at Zeppelin, with a higher relative BB contribution during the non-heating season. In summary, organic aerosol (281±106 ng m−3) constitutes a significant fraction of Arctic PM10, although surpassed by sea salt aerosol (682±46.9 ng m−3), mineral dust (613±368 ng m−3), and typically non-sea-salt sulfate SO42- (314±62.6 ng m−3), originating mainly from anthropogenic sources in winter and from natural sources in summer.

O artigo discute a corporeidade de avatares em plataformas digitais utilizados para a promoção de produtos de moda. O foco é analisar a presença digital da avatar Rennata, da marca Renner, e Pink, da influenciadora Bianca Andrade, conhecida como Boca Rosa. A questão norteadora é: os avatares digitais possuem corpos que dialogam com a representação da pluralidade e com as identidades dos sujeitos em ambientes online? Por ser um estudo inicial, a pesquisa caracteriza-se como Exploratória e Qualitativa, baseada em fontes bibliográficas e documentais. Nossa observação aponta para o processo de construção de uma corporeidade digital que transcende o domínio tecnológico dos ambientes virtuais e, por isso, é preciso considerar uma multiplicidade de questões sociais, políticas, individuais e de movimentação do mercado.

Campylobacter spp. é considerada a causa bacteriana mais comum de gastroenterite humana de origem alimentar no mundo, estando entre os “patógenos prioritários” resistentes a antimicrobianos que representam maior ameaça à saúde humana. A campilobacteriose humana é causada pelas espécies termofílicas de Campylobacter, sendo C. jejuni a principal espécie associada à doença em seres humanos e a mais comumente encontrada na cadeia produtiva de frango. Objetivou-se, com este estudo, detectar e enumerar Campylobacter spp. em carcaças resfriadas de frangos abatidos no oeste de Santa Catarina, Brasil. Foram analisadas 130 carcaças de frango coletadas após a etapa de pré-resfriamento por imersão em água pelo método ISO 10272-2:2017. Como resultado, não houve detecção de Campylobacter spp. nas amostras de carcaça de frango avaliadas durante o período de estudo e, consequentemente, não foi possível realizar a enumeração. Estes resultados podem ser atribuídos às limitações da metodologia utilizada, bem como à influência das medidas de biosseguridade adotadas pelas granjas avícolas integradas ao estabelecimento e dos procedimentos higiênico-sanitários e tecnológicos executados pelo estabelecimento no abate e processamento de aves através dos programas de autocontrole, que são essenciais para reduzir os riscos à saúde pública, tendo em vista a importância do estado de Santa Catarina na produção e exportação de carne de frango.

ABSTRACT In hydrological modeling and for the development of projects related to soil and water conservation and civil engineering, determination of water infiltration variables into the soil plays a key role. However, measuring infiltration with manual infiltrometers is work-intensive and requires several people, which casts doubts on the consistency of the process description. Our objective was to develop automated and compact systems for data acquisition by double-concentric-ring and Cornell infiltrometers. The systems are based on air differential pressure sensors and microcontrollers using open-source software and a simple construction. We developed ten sets of automatic infiltrometers for each method that were properly calibrated. The equipment was validated in the field, and the alignment of the automatically measured with the hand-measured infiltration data was considered satisfactory. The proposed systems make data records of infiltration and associated variables possible, with less operator dependence than manual measurement strategies. In addition, the enhanced resolution resulting from infiltrometer automation makes the infiltration curve more representative, especially in the initial infiltration stage.

Trust is an essential pre-condition for the acceptance of digital infrastructures and services. Transparency has been identified as one mechanism for increasing trustworthiness. Yet, it is difficult to assess to which extent and how exactly different aspects of transparency contribute to trust, or potentially impede it in cases of overwhelming complexity of the information provided. To address these issues, we performed two initial studies to help determining the factors that influence or have impact on trust, focusing on transparency across a range of elements associated with data, data infrastructures and virtual research environments. On one hand, we performed a survey among IT experts in the field of data science focusing on quality aspects in the context of re-using and sharing open source software, assessing issues such as the need for documentation, test cases, and accountability. On the other hand, we complemented this with a set of semi-structured interviews with senior researchers to address specific issues of the degree of transparency achievable with different approaches. They include, for example, the amount of transparency we can achieve with approaches from explainable AI, or the usefulness and limitations of data provenance in determining the suitability of data for reuse and others. Specifically, we consider mechanisms on three levels, i.e. technical, process-oriented as well as social mechanisms. Starting from attributes of trust in the “analogue world”, we aim to understand which of these can be applied in the digital world, how they differ, and what additional mechanisms need to be established, in order to support trust in complex socio-technological processes and their emergent results when the traditional approaches cannot be applied anymore.

Integrated crop-livestock systems are being increasingly used to intensify food production and make it more sustainable. On the other hand, most studies have focused on extensive systems. This paper analyzed the effects of different managements on soil and plants in an intensive integrated system for milk production. An experiment of management systems was installed in southern Brazil, Rio Grande do Sul, on a Latossolo Vermelho (Oxisol), in 2015 to evaluate: rotational grazing of dairy cows in the winter; rotational grazing of dairy cows in winter followed by soil chiseling; and ungrazed area (control). Soil physical properties and yields of crops were evaluated. Trampling by dairy cows increased soil bulk density by 24 % (0.0-0.05 m), but did not influence yields of subsequent soybean or maize. Chiseling reduced the bulk density of the uppermost layer by 19 %, but did not affect the yields of subsequent crops. It was concluded that in years with abundant water, dairy cattle grazing in an integrated crop-livestock system in Southern Brazil compacts the soil surface, but does not compromise the soil physical processes related to the growth and development of subsequent crops.

Abstract The inlet velocity of thermoplastic in injection molds plays a crucial role in obtaining high-quality polymer parts and the final performance of the product. It is known that the way the polymer is injected into the mold can directly affect important properties, such as the distribution of internal stresses, the cooling rate and the formation of surface defects. However, there are injection molding machines that only control injection pressure and dosage, making it difficult to obtain the gate inlet velocity into the mold cavity. Besides, some molds have many injection channels as well as complex inlet geometries, which make a challenging task to identify the inlet velocity. This study presents numerical and experimental approaches on how to determine the entry velocity in thermoplastic injection molds. The main results showed that these methods are highly efficient and contribute to identifying the gate inlet velocity with good accuracy.

Deep Learning has enabled significant progress towards more accurate predictions and is increasingly integrated into our everyday lives in real-world applications; this is true especially for Convolutional Neural Networks (CNNs) in the field of image analysis. Nevertheless, it has been shown that Deep Learning is vulnerable against well-crafted, small perturbations to the input, i.e., adversarial examples. Defending against such attacks is therefore crucial to ensure the proper functioning of these models—especially when autonomous decisions are taken in safety-critical applications, such as autonomous vehicles. In this work, shallow machine learning models, such as Logistic Regression and Support Vector Machine, are utilised as surrogates of a CNN based on the assumption that they would be differently affected by the minute modifications crafted for CNNs. We develop three detection strategies for adversarial examples by analysing differences in the prediction of the surrogate and the CNN model: namely, deviation in (i) the prediction, (ii) the distance of the predictions, and (iii) the confidence of the predictions. We consider three different feature spaces: raw images, extracted features, and the activations of the CNN model. Our evaluation shows that our methods achieve state-of-the-art performance compared to other approaches, such as Feature Squeezing, MagNet, PixelDefend, and Subset Scanning, on the MNIST, Fashion-MNIST, and CIFAR-10 datasets while being robust in the sense that they do not entirely fail against selected single attacks. Further, we evaluate our defence against an adaptive attacker in a grey-box setting.

Abstract Objective: This paper describes the first web-based self-completed 24-h recall designed to categorise food intake according to Nova groups – Nova24h – and its agreement with a reference tool in estimating the dietary relative contribution of the four Nova food groups (% of total energy intake). Design: Comparisons of estimates of dietary relative contributions of Nova groups obtained by Nova24h and one standard interviewer-led 24-h recall. Setting: Nationwide adult cohort study in Brazil. Participants: The subjects were 186 participants of the NutriNet Brasil Cohort Study (n 186). Results: No statistically significant differences were observed between the Nova24h and the reference tool mean contributions of unprocessed or minimally processed foods (52·3 % v. 52·6 %), processed culinary ingredients (11·6 % v. 11·9 %), processed foods (17·1 % v. 14·7 %) and ultra-processed foods (19·0 % v. 20·9 %). Intraclass correlation coefficients between individual estimates obtained for each Nova group showed moderate to good agreement (0·54–0·78). Substantial or almost perfect agreement between the tools was seen regarding the ability to rank participants according to quintiles of contribution of each Nova group (PABAK 0·69–0·81). Conclusions: Nova24h is a suitable tool for estimating the dietary relative energy contribution of Nova food groups in the NutriNet Brasil cohort. New studies are necessary to verify its adequacy in other populations.

Key Clinical Message The case highlights the good survival after radical surgery and chemotherapy of a cardiac sarcoma, and the need for close follow‐up due to possible early postsurgical complications.

ABSTRACT: There are no reported estimates of normal or abnormal bleeding in dogs during the hemilaminectomy and intervertebral disc fenestration procedures (HF), and that makes it difficult to study intraoperative complications, hemostasis methods, and the impact of bleeding on surgical performance. This prospective study estimated blood loss during HF procedures in dogs with intervertebral disc extrusion (IVDE). Blood loss was quantified by weighing gauzes during the surgical procedure, in addition to measuring the aspirated blood. Total blood volume was estimated according to body weight, and blood loss as a percentage of the total blood volume. The surgeon also subjectively classified the procedures, dividing them into intraoperative difficulty due to bleeding (G1) and absence of intraoperative difficulty (G2). Thirty-two dogs that underwent HF due to IVDE were included. The median blood loss (%) of dogs submitted to HF was 3.9 ± 9.67 of the total blood volume. There was a weak correlation between blood loss and weight, age, incision size, and procedure time.

Die Digitalisierung vollzieht sich auch in der Landwirtschaft in rasanter Geschwindigkeit. Die erheblichen Risiken bei der Anwendung von Lösungen, die in ihren Grundlagen und Auswirkungen oft nicht vollständig verstanden werden, bergen jedoch ein hohes Bedrohungspotenzial für die Resilienz und Nachhaltigkeit der Landwirtschaft. Dieser Artikel zeigt den Einsatz von Digitalen Zwillingen anhand von zwei ausgewählten Versuchsfarmen in Österreich. Das Projekt verfolgt das Ziel, durch den Aufbau modernster Versuchsfarmen als „Digitale Zwillinge“ eine zukunftsweisende Versuchsinfrastruktur für interdisziplinäre Forschung und Lehre auf internationalem Niveau zu etablieren. Es beinhaltet zudem ein Framework zur Integration von Anforderungen gemäß verschiedener Kriterien, um ein nachhaltiges Design von Digitalen Zwillingen in der Landwirtschaft zu ermöglichen.

Abstract Objective: This study evaluated the independent and combined environmental impacts of the consumption of beef and ultra-processed foods in Brazil. Design: Cross-sectional study. Setting: Brazil. Participants: We used food purchases data from a national household budget survey conducted between July 2017 and July 2018, representing all Brazilian households. Food purchases were converted into energy, carbon footprints and water footprints. Multiple linear regression models were used to assess the association between quintiles of beef and ultra-processed foods in total energy purchases and the environmental footprints, controlling for sociodemographic variables. Results: Both beef and ultra-processed foods had a significant linear association with carbon and water footprints (P < 0·01) in crude and adjusted models. In the crude upper quintile of beef purchases, carbon and water footprints were 47·7 % and 30·8 % higher, respectively, compared to the lower quintile. The upper quintile of ultra-processed food purchases showed carbon and water footprints 14·4 % and 22·8 % higher, respectively, than the lower quintile. The greatest reduction in environmental footprints would occur when both beef and ultra-processed food purchases are decreased, resulting in a 21·1 % reduction in carbon footprint and a 20·0 % reduction in water footprint. Conclusions: Although the environmental footprints associated with beef consumption are higher, dietary patterns with lower consumption of beef and ultra-processed foods combined showed the greatest reduction in carbon and water footprints in Brazil. The high consumption of beef and ultra-processed foods is harmful to human health, as well as to the environment; thus, their reduction is beneficial to both.

Este artigo apresenta o recorte de uma pesquisa em nível de mestrado que teve como objetivo compreender as contribuições para a aprendizagem durante o processo de produção de vídeos com tecnologias digitais por estudantes do Ensino Médio, para o VI Festival de Vídeos Digitais e Educação Matemática, com o tema educação financeira. Utilizamos o paradigma qualitativo, foram produzidos dados pela observação participante, notas de campo, questionário e grupo focal. Os participantes da pesquisa foram estudantes da turma do 1º ano “B”, da Escola Estadual Padre Arlindo Ignácio de Oliveira, em Campo Novo do Parecis - MT. A ferramenta de análise utilizada foram os Miniciclones de Transformações Expansivas. Os resultados do recorte de dados que apresentamos neste artigo indicam que com a produção dos vídeos os estudantes foram capazes de “fazer” matemática com autonomia, alterando o planejamento da professora em algumas situações.