Influence of data

E. de Noord, The influence of data preprocessing on the robustness nd parsimony of multivariate calibration models. Chemom. Intell. Lab. Systems, 23 (1994) 65-70,... 

Chen et al. (1998) developed an integrated approach, which can delete the influence of outliers in the data reconciliation problem, based on the idea of QQ-plots. The basic concept of this approach is to calibrate sampling data by means of its own main structure so that the influence of data decreases as it becomes less and less characteristic. In this way, the final data reconciliation procedure will be resistant to outliers. A limiting transformation, which operates on the data set, is defined to... 

The influence of data im and total photon count N = nm is exerted by adding the appropriate constraints to In P in proportions controlled by Lagrange multipliers Xm and p. The most likely solution hm is then found by maximizing... 

It is likely that an attacker has perfect knowledge about tlie original data for some part of the data set. In this case, the attacker simply replaces tlie watennarked data by the original data, thus erasing the watermark from the specific data elements. In general we found that reliable watermark detection can be acliieved even for a substitution of 80% of the watennarked data elements. However, this is only possible when many data elements are available at tlie decoder. Thus, it is worth to select for the watermarking process only data elements which are unlikely to be known by an attacker. The disturbing influence of data replacement can be prevented tliis way. 

Figure 3.14. The influence of data acquisition times on the ability to resolve fine structure. Longer acquisition times correspond to higher digitisation levels (smaller Hz/pt) which here enable characterisation of the coupling structure within the double-doublet (J = 6 and 2 Hz).

O.E. Denoord, The Influence of Data Preprocessing on the Robustness and Parsimony of Multivariate Calibration Models, Chemometrics and Intelligent Laboratory Systems, 23(1) (1994), 65-70. 

Well-managed data is fundamental to the dependability and operational integrity of a system. Many systems are not only reliant on data, but also the integrity of data. Therefore data should be addressed as part of the system safety case in common with other elements of the system. The system safety argument(s) should address the use of data and the influence of data errors on the system behaviour. However responsibility for data and its associated data integrity is often poorly defined. This lack of clarity allows vendors to abdicate responsibility for data, and its integrity to the client. 

The safety analysis should establish the nature and role of the data within the system and as a consequence, document the influence of data errors on operation of the system. This safety analysis should note that the lifetime of the data may exceed the specified working life of several generations of the implementation of the system [Needle 2003]. 

Data used by a safety-related system should be classified based upon the uses made of the data and the way in which the data influences the behaviour of the system. The nature and influences of data faults will also vaiy with the form and use of the data within a system. Data integrity requirements are essential if the suitability of data models is to be assessed. This paper has presented a process by which these data integrity requirements may be established. Additional design analysis may identify that the structure and composition of the data set or that data from the real world cannot be obtained in either the quantity, nor of the requisite quality. These data integrity requirements may also be used to identify verification and validation requirements for the system. 

Figure 3-52. Influence of (data point distance for point-point differentiation) on noise level R and SNR of a Gaussian band A = 100 mm FWHM = 16.65 mm) first derivative (according to [116]).

Niculescu, S. P., Kaiser, K. L. E., and Schuurmann, G. (1998) Influence of data preprocessing and kernel selection on probabilistic neural network modeling of the acute toxicity of chemicals to the fathead minnow and Vibrio fischeri bacteria. Water Qual. Res. J. Can. 33, 153-165. 

The coexistence of standard and safety programs in a F-CPU is possible because the safety data of the safety program are protected from unintended influences of data in the user program. 

What is readily apparent is that under the weighted PCA scheme the relative influence of data points that are considered as outliers should be low (i.e. small weight values). However, this assumes that the set of all weights are... 

---