No description available.

No description available.

No description available.

Available computing resources play a large part in enabling the training of modern deep neural networks to complete complex computer vision tasks. Improving the efficiency with which this computational power is utilized is highly important for enterprises to improve their networks rapidly. The first few training iterations over the data set often result in substantial gradients from seeing the samples and quick improvements in the network. At later stages, most of the training time is spent on samples that produce tiny gradient updates and are already properly handled. To make neural network training more efficient, researchers have used methods that give more attention to the samples that still produce relatively large gradient updates for the network. The methods used are called ''Importance Sampling''. When used, it reduces the variance in sampling and concentrates the training on the more informative examples. This thesis contributes to the studies on importance sampling by investigating its effectiveness in different contexts. In comparison to other studies, we more extensively examine image classification by exploring different network architectures over a wide range of parameter counts. Similar to earlier studies, we apply several ways of doing importance sampling across several datasets. While most previous research on importance sampling strategies applies it to image classification, our research aims at generalizing the results by applying it to object detection problems on top of image classification. Our research on image classification tasks conclusively suggests that importance sampling can speed up the training of deep neural networks. When performance in convergence is the vital metric, our importance sampling methods show mixed results. For the object detection tasks, preliminary experiments have been conducted. However, the findings lack enough data to demonstrate the effectiveness of importance sampling in object detection conclusively.

Estimating the variables of importance in inferentialmodelling is of significant interest in many fields of science,engineering, biology, medicine, finance and marketing. However,variable selection in high-dimensional data, where the number ofvariables is relatively large compared to the observed data points,is a major challenge and requires more research in order toenhance reliability and accuracy. In this bachelor thesis project,several known methods of variable selection, namely orthogonalmatching pursuit (OMP), ridge regression, lasso, adaptive lasso,elastic net, adaptive elastic net and multivariate adaptive regressionsplines (MARS) were implemented on a high-dimensional dataset.The aim of this bachelor thesis project was to analyze andcompare these variable selection methods. Furthermore theirperformance on the same data set but extended, with the numberof variables and observations being of similar size, were analyzedand compared as well. This was done by generating models forthe different variable selection methods using built-in packagesin R and coding in MATLAB. The models were then used topredict the observations, and these estimations were compared tothe real observations. The performances of the different variableselection methods were analyzed utilizing different evaluationmethods. It could be concluded that some of the variable selectionmethods provided more accurate models for the implementedhigh-dimensional data set than others. Elastic net, for example,was one of the methods that performed better. Additionally, thecombination of final models could provide further insight in whatvariables that are crucial for the observations in the given dataset, where, for example, variable 112 and 23 appeared to be ofimportance. / Att skatta vilka variabler som är viktigai inferentiell modellering är av stort intresse inom mångaforskningsområden, industrier, biologi, medicin, ekonomi ochmarknadsföring. Variabel-selektion i högdimensionella data, därantalet variabler är relativt stort jämfört med antalet observeradedatapunkter, är emellertid en stor utmaning och krävermer forskning för att öka trovärdigheten och noggrannheteni resultaten. I detta projekt implementerades ett flertal kändavariabel-selektions-metoder, nämligen orthogonal matching pursuit(OMP), ridge regression, lasso, elastic net, adaptive lasso,adaptive elastic net och multivariate adaptive regression splines(MARS), på ett högdimensionellt data-set. Syftet med dettakandidat-examensarbete var att analysera och jämföra resultatenav dessa metoder. Vidare analyserades och jämfördes metodernasresultat på samma data-set, fast utökat, med antalet variableroch observationer ungefär lika stora. Detta gjordes genom attgenerera modeller för de olika variabel-selektions-metodernavia inbygga paket i R och programmering i MATLAB. Dessamodeller användes sedan för att prediktera observationer, ochestimeringarna jämfördes därefter med de verkliga observationerna.Resultaten av de olika variabel-selektions-metodernaanalyserades sedan med hjälp av ett flertal evaluerings-metoder.Det kunde fastställas att vissa av de implementerade variabelselektions-metoderna gav mer relevanta modeller för datanän andra. Exempelvis var elastic net en av metoderna sompresterade bättre. Dessutom drogs slutsatsen att kombineringav resultaten av de slutgiltiga modellerna kunde ge en djupareinsikt i vilka variabler som är viktiga för observationerna, där,till exempel, variabel 112 och 23 tycktes ha betydelse. / Kandidatexjobb i elektroteknik 2021, KTH, Stockholm

Interruptions in power supply are inevitable due to faults in power system distribution network. These interruptions are not only expensive for the customers but also for the distribution system operator in the form of penalties. Increase in system redundancy or the use of component-specific sensors can help in reduction of interruptions. However, these options are not always economically feasible. Therefore, there is a need to check for other possibilities to reduce the risk of outages. The data stored in substations can be used for reducing the risk of outages by deriving component importance indices followed by ranking and predicting the outages. This thesis presents component importance indices derived by identifying the critical components in the grid and assigning index based on certain criterion. The model for predicting the faults is based on the weather conditions observed during the outages in the past. Component importance indices are derived and ranked based on the de-energisation time of components, frequency and impact of outages. This helps prioritize components according to the chosen criterion and adapt monitoring strategies by focusing on the most critical components. Based on categorical Naive Bayes, a model is developed to predict the probability of fault/failure, location and component type likely to be affected for a given set of weather conditions. The results from the component importance indices reveal that each component’s rank varies based on the chosen criterion. This indicates that certain components are critical with respect to specific criterion and not all criteria. However, some components are ranked high in all the methods. These components are critical and need focused monitoring. The reliability of results from component importance indices to a great extent depends on the time frame of the outage data considered for analysis. The prediction model can alert the distribution system operator regarding the possible outages in the network for a given set of weather conditions. However, the prediction of location and component type likely to be affected is relatively inaccurate, since the number of outages considered in the time frame is low. By updating the model regularly with new data, the predictions would be more accurate. / Avbrott i strömförsörjningen är oundvikliga på grund av fel i distributionsnätet för kraftsystemet. Dessa avbrott är inte bara dyra för kunderna utan också för distributionssystemoperatören i form av påföljder. Ökad systemredundans eller användning av komponentspecifika sensorer kan hjälpa till att minska avbrott. Dessa alternativ är dock inte alltid ekonomiskt genomförbara. Därför är det nödvändigt att kontrollera om det finns andra möjligheter för att minska risken för avbrott. Data lagrade i transformatorstationer kan användas för att minska risken för avbrott genom att härleda komponentviktindex följt av rangordning och förutsäga avbrott. I denna avhandling härleds viktighetsindex genom att identifiera de kritiska komponenterna i nätet och tilldela index baserat på vissa kriterier. Felprognoserna gjordes baserat på de väderförhållanden som observerades under avbrott. komponentviktighetsindex härleds och rankas baserat på komponenternas urladdningstid, frekvens och påverkan av avbrott. Detta hjälper till att prioritera komponenter enligt det valda kriteriet och anpassa övervakningsstrategier genom att fokusera på de mest kritiska komponenterna. Baserat på kategoriska Naive Bayes utvecklas en modell för att förutsäga sannolikheten för fel / fel, plats och komponenttyp som sannolikt kommer att påverkas under en viss uppsättning väderförhållanden. Resultaten från komponentviktighetsindexen visar att varje komponents rang varierar beroende på det valda kriteriet. Vissa komponenter rankas dock högt i alla metoder. Dessa komponenter är kritiska och behöver fokuserad övervakning. Tillförlitligheten hos resultat från komponentviktindex beror till stor del på tidsramen för avbrottsdata som beaktas för analys. Prognosmodellen kan varna distributionssystemoperatören om möjliga avbrott i nätverket för en viss uppsättning väderförhållanden. Förutsägelsen av plats och komponenttyp som sannolikt kommer att påverkas är dock relativt felaktig, eftersom antalet avbrott som beaktas i tidsramen är lågt. Genom att uppdatera modellen regelbundet med nya data skulle förutsägelserna vara mer exakta.

This study investigated how social media use is related to body esteem (particularly weight satisfaction and muscularity importance) in adolescents. We conducted a 17-day Ecological Momentary Assessment study among 183 adolescents (12–17 years, 58% girls). Each adolescent reported on his/her social media use, weight satisfaction, and muscularity importance four times per day (68 assessments per participant; 6,863 completed in total). Using a person-specific, N=1 method of analysis (Dynamic Structural Equation Modeling), we found that at the between-person level, social media use is not associated with lower satisfaction with one’s weight and greater importance of looking masculine. At the within-person level, we found a significantly negative association for weight satisfaction and a significantly positive association for muscularity importance. For weight satisfaction, 76% of adolescents experienced no or very small effects as a result of SMU, 2% experienced positive effects, and 22% experienced negative effects. Regarding muscularity importance, 89% experienced no or very small effects of SMU on muscularity importance, 10% experienced positive effects, and 1% experienced negative effects. There is little evidence of gender differences in the effect of social media on body esteem in our sample.

The purpose of this paper is to investigate what pupils who study Spanish as a foreignlanguage in the senior level of the nine year compulsory school, think about how theylearn to talk Spanish. What is their opinion about what it takes to be able to speak andcommunicate in Spanish? And what type of exercises do they prefer? This paper alsoaims at investigating if the pupils understand the importance of reading, writing andlistening to as much Spanish as possible if they want to be good at talking Spanish. Inour investigation are we using the Common European Framework of Reference forlanguages: Learning, Teaching, Assessment (CEFR) and other theories that supporttheir conclusions. A quantitative method is applied, a questionnaire is given to 108students in the seventh, eighth and ninth grade in a school in the southern part ofSweden.The results of the questionnaire have shown that when it comes to learn to talkSpanish the pupils believe most in practicing talking in Spanish. 72% of the pupils haveanswered that they agree completely with this assertion. They also understand theimportance of the teacher speaking Spanish during class, 58% completely agree that thisis important. Talking Spanish is classed as output, and listening to the teacher talkingSpanish goes under the term input. The type of activity that they prefer when they talkSpanish is to talk in small groups with friends/classmates, 38% of the pupils think thatthis is the best method. The activity that they prefer the least is to make presentations infront of the class (11%). We have also calculated by using Fisher´s exact test if there isa connection between how they have answered the questions and their age and gender.In only one case was there a connection with statistical certainty. The test showed thatwhen it comes to speaking when everyone is listening, the girls are those whoexperience it hardest to do.

Portfolio Management is the process of allocating an investor's wealth to in­ vestment opportunities over a given planning period. Not only should Portfolio Management be treated within a multi-period framework, but one should also take into consideration the stochastic nature of related parameters. After a short review of key concepts from Finance Theory, e.g. utility function, risk attitude, Value-at-rusk estimation methods, a.nd mean-variance efficiency, this work describes a framework for the formulation of the Portfolio Management problem in a Stochastic Programming setting. Classical solution techniques for the resolution of the resulting Stochastic Programs (e.g. L-shaped Decompo­ sition, Approximation of the probability function) are presented. These are discussed within both the two-stage and the multi-stage case with a special em­ phasis on the former. A description of how Importance Sampling and EVPI are used to improve the efficiency of classical methods is presented. Postoptimality Analysis, a sensitivity analysis method, is also described. / Statistics / M. Sc. (Operations Research)

Thesis (MDevF (Business Management))--University of Stellenbosch, 2009. / ENGLISH ABSTRACT: The importance of market microstructure in determining the success of a bond market in allocating financial resources depends on the degree to which the microstructure elements like liquidity, efficiency and volatility have been designed to determine the proper price at which matching of demand and supply in an efficient and effective manner is done. This research project analyzes some of the fundamental microstructure elements responsible for the current state of the Botswana bond market. The Botswana bond market is still in its infant stage hence there is little information on trades, which contributes to the liquidity problem. The purpose of the study was to investigate the liquidity and efficiency in Botswana’s bond market. The study also sought to compare the behaviour of the Botswana bond market to those of South Africa and further indicate what is behind the bond market emergence. Houweling, Mentink and Vorst‘s (2003) measure was used, in addition to a combination of simple regression and latent models. In the test of efficiency, a static model has been employed. Overall, it is established that the corporate bond market is less efficient and is illiquid. Furthermore, it is revealed that Botswana is still lagging behind South Africa when it comes to the level of development of the corporate bond market.