Distance vs Sample Clauses

Distance vs. Because bigger δ will result in less candidates in δ-location set, the distance declines when δ increases. Figures 3.5f, 3.6f and 3.5f show that PIM achieves better accuracy than LM. However, from 10−1.5 to 10−1, the improvement on distance is very small while privacy guarantee drops signifi- cantly as in Figures 3.5d, 3.6d and 3.7d. Especially in Figure 3.6d, size of ∆X is 1 when δ = 0.1. Therefore, choosing a high value of δ (like δ > 0.03) does not provide the best trade-off of privacy and utility. Impact of Markov model. Comparing Figures 3.5 on popular M and Figure