### Table 1: Comparison of algorithmic results for compressed sensing

"... In PAGE 3: ... Our algorithm achieves a uniform bound because, unlike for each algorithms, HHS uses a stronger estimation matrix and a combination of sifting and noise reduction matrices (see below) tailored to the mixed- norm bound of Theorem 1. (We include in Table1 uniform results only.) Chaining Pursuit is the only algorithm in the literature that achieves the first three desiderata [10].... ..."

### Table 1: Comparison of algorithmic results for compressed sensing

"... In PAGE 3: ... Our algorithm achieves a uniform bound because, unlike for each algorithms, HHS uses a stronger estimation matrix and a combination of sifting and noise reduction matrices (see below) tailored to the mixed- norm bound of Theorem 1. (We include in Table1 uniform results only.) Chaining Pursuit is the only algorithm in the literature that achieves the first three desiderata [10].... ..."

### Table 1 Framework for Scenario Modeling

"... In PAGE 24: ... We use the terms not in their absolu te sense but in their relative relationship. Table1 presents a scheme for planning a series of simulations, where three separate scenarios of exogenous ... In PAGE 25: ...Table1... ..."

### Table 4: Computing times for various steps of our compression and re- construction framework.

### Table 1. Performance of lossy image compression in JPEG 2000 framework (PSNR, dB)

"... In PAGE 8: ... Table1 shows that our color space transform LAR performs better for all the im- ages at all the bit rates, average 4.3dB higher, minimum 3dB and maximum 6dB better than the default color transform of JPEG 2000.... ..."

### Table 4. The effect of raw sensing data filtering.

"... In PAGE 14: ... At stage 3, the number of symbols is reduced by 90% and the final area sequence is fed to grammar hierarchy. Due to the sensor calibration the number of symbols eventually fed as input to the grammar hierarchy (approximately 10 to 100) are orders of magnitude less than the initial number of image centroids recorded (2583 to 6648), as shown in Table4 . These numbers demonstrate the feasibility of such a system running in real time on a sensor network.... In PAGE 14: ... In addition, the fact that activities lasting as much as 50 minutes can be reduced down to a sequence of only 100 symbols shows that modeling human activity as a sequence of actions could meet the real time requirements and limitations of sensor networks. Table4 provides some more insight into the effect of sensor calibration in the size of input fed to the inference framework. In the case of dinner preparation, 6648 image locations were acquired that were finally reduced down to only 109 area symbols.... ..."

### Table 3: Framework Dimensions for MOMIS and KRAFT

"... In PAGE 8: ...a0 a2 a11 encode AU SENSUS semantic validation store LCG KW enter data resources Database Classifier sense graph Lexical Interface retrieve Matcher represent query answer adapt to Spanish author KIF-Number Ontology AU AD Chemicals Ontology Ontology KIF-List Ontology OA Chemical-Elements Ontology GUM Standard Units Ontology Chemical-Crystals Ontology Standard- Dimensions Ontology Ontogeneration Integrate KPML Table3 : Framework Dimensions for MOMIS and KRAFT MOMIS KRAFT Maturity Level An advanced prototype. Wrappers and mediators are quite mature, and are used commercially in data warehouses.... ..."

### Table 1: Algorithm frameworks for the solution of [VIP]

"... In PAGE 5: ...in this case, the step length rule will always be R. To conclude, an instance of the cost approximation algorithm is described by the eight-tuple ([X; F; u]; [ k; MSU; EA]; [ kC; EAR]): (3) 4 Relations between algorithm frameworks In Table1 , we provide a list of several existing algorithm frameworks for the solution of [VIP]. Each entry corresponds to an algorithm class, for which there are known convergence results.... In PAGE 5: ... The descriptions are derived from relations between these algorithms and the cost approximation algorithm investigated in [22, 23, 24, 25]. 6 Remarks An immediate result of the classi cation is that we nd, from Table1 , that most of the algorithm frameworks that have been developed for the solution of [VIP] and its special cases have been established both with stronger assumptions than necessary and with weaker results than possible and that have been obtained elsewhere. (The monotonicity requirements on F are, according to the analysis made in [24, 25, 26, 27], only implicit, in the sense that the requirements are given through the dependency on F of the cost approximation mapping corresponding to a certain algorithm instance; in the analysis made for many of the algorithm frameworks given in Table 1 the mapping F is, however, a priori supplied with a monotonicity property which is not always necessary.... ..."