Exploration in non-stationary bandits | The actual reward earned on an individual trial was given by a sample from a Gaussian distribution with the current mean, and a standard deviation of 4. |
Exploration in non-stationary bandits | In addition, when the discount parameter is larger (7 = 0.90 vs 7 = 0.99) the algorithm samples more often, consistent with the larger difference in utility for a given standard deviation for larger discount parameters (Fig. |
Information sampling | Standard deviation option 1 |
Information sampling | Utility as a function of mean of option 1 and mean of option 2, with standard deviation of both options set to 4 and discount rate, v = 0.90. |
Information sampling | Utility as a function of standard deviation for 2 discount values when mean is 50 for both options and standard deviation is 4 for option 2. |
Markov decision processes | For the non-stationary two-armed bandit, the means were fit with a 3rd order B-spline, and the standard deviations were fit with a 2nd order piecewise polynomial. |
Markov decision processes | The node locations for the standard deviations were given by 0.25, 1, 3, 5, and 15. |
Markov decision processes | The standard deviations were evaluated at 0.5, 1, 2, 3, 4, 5, 7 and 14. |
Task specific details of the MPD models | Therefore the information space for this POMDP, equivalent to the state space for the MDP, can be compactly represented: bt = ([11, 61, [12, 62) as the estimated mean and standard deviation (or variance) of each bandit. |
Task specific details of the MPD models | This integral was calculated numerically by discretizing y over +/- 2 standard deviations , and sampling 10 points. |
Microarray metadataset | Instead, we found that the best approach was to convert the values in each time series to Z-scores—i.e., for each gene in each dataset, we subtract its mean eXpression level and divide by its standard deviation . |
Microarray metadataset | To prevent the need to recalculate the null distribution for every pattern of NAs in the data for empirical ITK_CYCLE (there were 5005 unique NA permutations in the data), the NAs were replaced by random noise drawn from a Gaussian distribution with mean and standard deviation that match those of the data on the whole. |
Simulated data benchmarks | 3A we generated 10,000 time series with uniformly distributed random phase shifts (always with a 24 h period) and added Gaussian noise to each point with a standard deviation of 25% or 50% of the total waveform amplitude, examples of which can be seen in Fig. |
Simulated data benchmarks | (B) Cosine in black, with Gaussian noise with standard deviation of 25% (blue) or 50% (green) of amplitude. |
Simulated data benchmarks | We added Gaussian noise with a standard deviation of either 25% or 50% of the amplitude of the time series, as previously described. |
Supporting Information | AUROCs for simulated data with 25% noise ( standard deviation of Gaussian noise as a percent of amplitude). |
Detection of artificial EPSCs immersed in fluctuating noise | We injected into layer 2/3 pyramidal cells current consisting of three components: simulated, artificial excitatory postsynaptic current (aEPSC) from a single presynaptic neuron, fluctuating noise with standard deviation 0, and a DC offset (Fig. |
Detection of artificial EPSCs immersed in fluctuating noise | After this adjustment, the overall standard deviation of the scaled fluctuating current awas between 70 and 110pA. |
Experiment 1. Partially-defined input: Artificial EPSCs immersed in fluctuating noise | Current for injection in this first set of experiments was composed of 1) a fluctuating component 017(t), where 17(t) is a standardized (zero mean, unit variance) Ornstein-Uhlenbeck (OU) process with a correlation time of T = 5ms rescaled to have standard deviation 0, 2) artificial EPSCs of several different amplitudes: 0.1, 0.2, 0.25, 0.3, 0.5, 1.0 and 1.5 of the noise standard deviation 0, and 3) a DC component tuned to maintain a desired firing rate, around ~ 5 Hz. |
U | Error bars denote standard deviation across cells. |
U | Error bands denote standard deviation across cells. |
Data segmentation | In addition Gaussian fit of the resulting delay histograms for values 0.85 g VC 3 0.95 did not show significant differences either in the standard deviation or the 95% confidence interval from a Gaussian fit of Fig. |
Results | Breakdown of the original spatio-temporal data into the individual and paired flight trajectories together with mean and standard deviation values of the temporal and spatial length as well as speed of all flights. |
Supporting Information | Computing the mean and standard deviations we obtain 349.4° and 169.4°, (i33.1°), for the individual flying bats and 348.0° and 168.0°, (i36.7°), for the paired flying bats. |
Network-based prediction of DrugBank ligand and human target pairs | The “Global Z-score” (G2) is obtained by running the predictions of all drugs present in DrugBank against all targets, obtaining a global mean (1m) and a global standard deviation (CG) to Z-score a specific predicted pair. |
Network-based prediction of DrugBank ligand and human target pairs | The “Local Z-score” (L2), is similarly calculated by running the predictions of all drugs present in DrugBank retrieving the mean (1,4 L) and the standard deviation (CL) of the score for a specific target. |
nAnnoLyze benchmarking | Mean values and standard deviation after 10-fold cross-validation. |
Stimuli and task | The standard deviation of the Gaussian blob was only 3.5 mm (0.37 deg) on the screen in order to provide visually very reliable feedback. |
inter Iation test sha es '—' extrapolation W p extrapolation test shape shape parameter p test shape | Black circles and error bars indicate the mean and standard deviation across participants. |
inter Iation test sha es '—' extrapolation W p extrapolation test shape shape parameter p test shape | The yellow line and shaded area show the Mixture-of-Kalman-Filters Model mean and standard deviation across participants. |