Alexander Tschantz (8755416), Anil K. Seth (8755419), Christopher L. Buckley (8755422) · Figshare · 2020
<p><b>(A) Number of pruned parameters.</b> Percentage of times each parameter was pruned, averaged over 300 agents. A parameter was pruned if it <i>decreased</i> the evidence for agents model. <b>(B) Total pruned parameters.</b> The average number of total number of pruned parameters, averaged over 300 agents.</p>
Cited by 0Open access
J. A. Marmolejo (3778969), Jonás Velasco (3778972), Héctor J. Selley (3778966) · Figshare · 2017
Cited by 0Open access
2012
fail to reproduce yields, particularly at farm, regional and larger scales. This is largely due to inadequate cific field experiments performed in one or few years, such models have also been used to extrapolate model results in time and space – most commonly done by aggregating inputs or outputs. More recent attempts have tried to integrate these models or their out-puts, i.e. simulated yields, into higher level models representing farms (e.g. Matthews et al., 2006) or regions (e.g. Gimona et al., 2006). However, it has also been shown, that plant growth models et al., 2008). Indeed, in most plant growth models, the system sim-ulated is seen as a set of biophysical processes that can be controlled with unlimited resources (labour, machinery) and instantaneous operations assuming that the farmer can intervene on his fields whenever and wherever he wants. In grasslands, such modelling assumption may lead to an overestimation of harvested yields. In-deed, a delay of harvesting time due e.g. to weather constraints or lack of machinery available can result in yield reduction and addi-tionally to losses of forage nutritive value (Fig. 1). A promising approach to overcome these limitations and to pro-vide more accurate management information is to link plant
Cited by 0
Christian S. Göbl (330948), Latife Bozkurt (330947), Andrea Tura (297965) et al. · Figshare · 2015
<p>Data represent the estimated (mean) number of included variables in m = 5000 simulations for n = 50, 110 and 500 with 9 and 5 meaningful variables and 0 and 4 nuisance variables for scenario B and C, respectively. For scenario C the average numbers of included nuisance variables are given in parentheses. Ordinary least square estimation (OLS); Stepwise-backward variable selection with AIC (sAIC) and BIC (sBIC); Penalized estimation by using the Ridge method (Ridge), the least absolute shrinkage and selection operator (LASSO) and the elastic net L1-L2 (Enet 1), elastic net L2-L1 (Enet 2), elastic net L1+L2 (Enet 3).</p><p>Model complexity.</p>
Cited by 0Open access
Albert Oosterhof, Faranak Rohani, Peggy Stillwell et al. · 2008
design of multimedia instructional materials and customized assessments, CALA assists policy makers and educators by providing them analyses and practical solutions.
Cited by 0
P.W. Bogaart, L.M.W. Akkermans · 2009
Cited by 0
D.J.J. Walvoort, G.A.K. van Voorn · 2012
Cited by 0
Arthur Charpentier · OpenEdition (OpenEdition) · 2013
Today, Olivier Scaillet gave a great talk on fast recursive projections. The idea was great, and the talk was amazing. A great plenary session talk actually. And after lunch, while we were having a coffee, we started to discuss about financial model complexity, mentioning that sometimes, traders and quants are lost, and it might be good to spend more time on basics than on very advanced stuff (which was, in fact, the general idea of my own talk, yesterday). And Olivier recalled that story, on...
Cited by 0Open access
Oliver Nelles · 2001
Cited by 3
BATTAGLIESE, Diego · IRIS Research product catalog (Sapienza University of Rome) · 2020
In the present thesis, we describe and explore a new method of constructing prior distributions on the additional model components that build up a more flexible model starting from a base model which would not include those components. First of all, we want to briefly sketch the idea of the PC prior. Suppose to have a certain model that could be rendered more flexible and richer by introducing an extra-component. As a toy example, suppose one wants to make more robust the Gaussian distribution by allowing an additional parameter to control for kurtosis. In this specific case, we are dealing with a Student-t distribution, so a prior for the degrees of freedom needs to be specified. The concept of PC prior arises from the distance between the simpler model and the more flexible one. The PC prior is closely related to the Kullback-Leibler divergence (KLD), that is meant to penalise deviations from the base model. Afterwards, the KLD is transformed into a more interpretable distance scale, and such a distance is supposed to be exponentially distributed. The selection of the rate parameter of the exponential distribution is left to the user belief and, finally, by performing a change of variable from the distance scale to the parameter of interest the PC prior is obtained.
Cited by 1Open access
Pascal Marijan, Sebastian Igel, Tatjana Legler et al. · AIP conference proceedings · 2026
Cited by 0
T. A. Hauge, B. Lie · 2015
A 528 order mechanistic model of a paper machine is implemented in Mat-lab. The model has been developed for advanced control of three key quality variables, and it is desirable to reduce the size and complexity of the full scale model. It is shown how the full scale model can be reduced by both system identification techniques and by utilizing our physical knowledge about the pro-cess. The prediction abilities of the various reduced order models are compared with the output from the 528 order model, highlighting some distinct features of the various models.
Cited by 0
Toby Dylan Hocking, Gudrun Schleiermacher, Isabelle Janoueix-lerosey et al. · 2013
A map of genomic copy number alterations in neuroblastoma based on annotation-guided breakpoint detection
Cited by 0
Raúl Espejo · Advances in information quality and management · 2018
In this chapter, the author proposes an enterprise complexity model (ECM), which is visualized as a methodology to achieve the distributed governance of an ecology of evolving enterprises. Governance is understood as guiding the enterprises self-organization towards policies creating, regulating, and producing products and services for society. Self-organization is grounded in the communications and interactions of stakeholders. The purpose of an ECM model is not institutional development but guiding, enabling and facilitating interactions of all kinds with the support of current and disruptive technologies to increase society's requisite variety to deal with social, ecological, and economic challenges. An enabling context helps the branching of the enterprises' creativity into all kinds of innovations, forms of coordination, and operational alignment of their interests. Quality of life, fairness, and social justice are the values driving this ecology of enterprises towards a deeper and wider appreciation of issues of social concern.
Cited by 3
Daniel Durstewitz · Bernstein series in computational neuroscience · 2017
Cited by 1
Stefano Rovetta, Francesco Masulli, Alberto Cabri · Lecture notes in computer science · 2017
Cited by 3
Chris McConnell · Lecture notes in computer science · 1995
Cited by 0
Ivar L. Thorson (838915), Jean Liénard (5666239), Stephen V. David (185849) · Figshare · 2015
<p>Pareto plots compare model parameter count (horizontal axis) versus prediction correlation (vertical axis) for each linear STRF architecture, averaged over <i>N</i> = 176 A1 neurons. The Pareto front (red line) indicates the best prediction correlation for models with parameter count at or below the current abscissa. Dark gray points indicate models at the focus of this study, varying in only in the linear filter module and sharing the same input compression, spike nonlinearity, and fit algorithm. Black, purple, and orange points/arrows indicate the FIR, factorized, and parameterized models, respectively, explored in detail in later sections. <b>A.</b> For estimation data, prediction correlation tends to increase for more complex models. <b>B.</b> For validation data, performance reaches its maximum at just 29 parameters, suggesting that the increase in the estimation data for higher parameter counts reflects overfitting to noise by the more complex models. <b>C.</b> Summary of model architecture variants. Each bar shows the number of architectures evaluated using modules and fitting algorithms that differed from the core set of modules and procedures detailed in subsequent figures. For example, “Filterbank+output NL” indicates the number of models tested with a filterbank other than the second-order gammatone (<a href="http://www.ploscompbiol.org/article/info:doi/10.1371/journal.pcbi.1004628#pcbi.1004628.e013" target="_blank">Eq 11</a>) and output nonlinearity other than the double exponential sigmoid (<a href="http://www.ploscompbiol.org/article/info:doi/10.1371/journal.pcbi.1004628#pcbi.1004628.e016" target="_blank">Eq 14</a>).</p>
Cited by 0Open access
William R. Holmes (291716), Leah Edelstein-Keshet (141838) · Figshare · 2015
<p>A summary of the models on a two-axes “complexity” plane. The horizontal axis represents the level of biological detail (increasing from left to right) and the vertical axis represents the computational complexity of the simulation (increasing from bottom to top). See <a href="http://www.ploscompbiol.org/article/info:doi/10.1371/journal.pcbi.1002793#pcbi-1002793-t001" target="_blank">Table 1</a> for a listing of specific biological and computational aspects of these models.</p>
Cited by 0Open access
Norman F. Schneidewind · International Journal of Reliability Quality and Safety Engineering · 2008
A model of software complexity and reliability is developed that uses an evolutionary process to transition from one software system to the next while complexity metrics are used to predict the reliability for each system. Systems are tested until the software passes defect presence criteria and is released. Testing criteria are based on defect count, defect density, and testing efficiency predictions exceeding specified thresholds. In addition, another type of testing efficiency — a directed graph representing the complexity of the software and defects embedded in the code — is used to evaluate the efficiency of defect detection in NASA satellite system software. Complexity metrics were found to be good predictors of defects and testing efficiency in this evolutionary process.
Cited by 2