diff options
Diffstat (limited to 'paper/rebuttal.txt')
| -rw-r--r-- | paper/rebuttal.txt | 153 |
1 files changed, 81 insertions, 72 deletions
diff --git a/paper/rebuttal.txt b/paper/rebuttal.txt index 2b7d32b..25a9e62 100644 --- a/paper/rebuttal.txt +++ b/paper/rebuttal.txt @@ -1,90 +1,59 @@ -""" -multiple sources don't make much of a difference in their model, because -[...]. So if two -cascades originate at sources that are more than a constant distance away from -each other, it's the same as two consecutive, independent cascades. -""" +R2: +" +The set of parameters \theta always lies in some constrained space. For example, +in the independent cascade model, \theta_{i,j} < 0; in the voter model, +\sum_{i,j} \theta = 1 and \theta_{i,i} \neq 0.[...] If the authors incorporate +the constraints on the model parameters, it is not clear whether [Neghahba, +2012] will be applicable anymore since it assumes decomposable regularizers. +However, the authors would have (norm_1 + regularization induced by +constraints), which is not really clear whether is decomposable or not. +" +ANS:TODO -ANS: This is an interesting point. However, in the problem we study the graph is -unknown to us. Suppose that two cascades start at the same time at two very -different points in the graph. Despite the fact that the infected nodes from -each cascade will not overlap, we cannot in practice attribute an infected node -to either cascade because this information is hidden to us. - - -"""In the independent cascade model, nodes have one chance to infect their +" +In the independent cascade model, nodes have one chance to infect their neighbors. However, the definition in section 2.2.1. seems to allow for multiple attempts, since at any given time t+1, the probability depends on \theta_j X_t -""" -ANS: As the reviewer correctly points out, the standard ICC model does not allow +" +As the reviewer correctly points out, the standard ICC model does not allow for multiple infection attempts over time. The definition of section 2.2.1 also prohibits multiple attempts by considering that nodes stay active for only one time step, defining X_t as the set of nodes active at the previous time step only, and saying that only nodes which have not been infected before are susceptible to be infected. - -"""what would be the guaranteed/expected performance given some number of -cascades? -""" -ANS: This is an interesting point. For the experiment section, we could -calculate the theoretical guarantees for the synthetic graphs and observe -whether or not the theoretical bounds are pessimistic in practice. - - -"""Where is the explanation about Figure 1(f)? What is p_init? -""" -ANS: This is a typo. It should read "n" the number of cascades. - - -""" -Running time is not discussed here. -""" -ANS: This is a valid point. The MLE algorithm from Netrap.-Sangh. has similar -running time to the penalized MLE algorithm. Their greedy algorithm runs -considerably faster at the price of a slower convergence rate in practice. A -precise comparison of running times can be be included. - - -""" -Citations/Related work remarks -""" -ANS: the corresponding requested citations can be included on lines 42, 68, 75, -78, 93, 362. The authors regret not to have cited Du et al. 2012 and their work -should be included in the related work section along with other work considering -the estimation of influence in networks. It can be mentioned that Daneshmand et -al adopt the same model as Gomez-R et al '10 and Abrahao et al. '13. The -phrasing can be changed from "Graph Inference" to "Network Inference" with the -requested citations. - -""" +" in a continuous-time model there are not necessarily steps and property 1 is not necessarily satisfied -""" -ANS: This is a good point and the distinction can be made. - - -""" -the unrealistic assumption of one time infection chance. -"" -ANS: it is true that the standard discrete-time independent cascade model -studied by Netrapalli-Sanghavi assumes one time infection chance but this -restriction is not made at the Generalized Linear Cascade level and is specific -only to the example of section 2.2.1 +" +This is a good point and the distinction can be made. +R3: +" +multiple sources don't make much of a difference in their model, because [...]. +So if two cascades originate at sources that are more than a constant distance +away from each other, it's the same as two consecutive, independent cascades. +" +This is an interesting point. However, in the problem we study the graph is +unknown to us. Suppose that two cascades start at the same time at two very +different points in the graph. Despite the fact that the infected nodes from +each cascade will not overlap, we cannot in practice attribute an infected node +to either cascade because this information is hidden to us. -""" -The authors need to show at least one common metric for all types for graphs -"" -ANS: this is an interesting point: A graph plotting the same metric for all -considered networks can replace one of the 6 figures. +" +Running time is not discussed here. +" +This is a valid point. The MLE algorithm from Netrap.-Sangh. has similar running +time to the penalized MLE algorithm. Their greedy algorithm runs considerably +faster at the price of a slower convergence rate in practice. A precise +comparison of running times can be be included. -""" -the inference in discrete time, one-time-susceptible contagion +" +The inference in discrete time, one-time-susceptible contagion processes is less interesting and easier than the continuos version. -""" -ANS: This is a valid point. Some things to note are: the generalized cascade +" +This is an interesting point. Some things to note are: the generalized cascade model class is sufficiently flexible to include multiple-time-susceptible contagion processes (such as the linear voter model). Furthermore, it is not immediately clear that discrete-time processes cannot approximate some @@ -93,5 +62,45 @@ continuous time process with exponential transmission likelihood by considering intervals of time of length dt, binning infections to these intervals, and considering that nodes remain infected until the final observation time. By exploiting the memoryless property of the exponential distribution, we recover -its discrete-time analog: the geometric distribution for dt<<1. The problem is +its discrete-time analog: the geometric distribution when dt<<1. The problem is still decomposable and fits into the Generalized Linear Cascade model framework. + +" +the unrealistic assumption of one time infection chance. +" +It is true that the standard discrete-time independent cascade model +studied by Netrapalli-Sanghavi assumes one time infection chance but this +restriction is not made at the Generalized Linear Cascade level and is specific +only to the example of section 2.2.1 + +R4: +" +what would be the guaranteed/expected performance given some number of +cascades? +" +This is an interesting point. For the experiment section, we could +calculate the theoretical guarantees for the synthetic graphs and observe +whether or not the theoretical bounds are pessimistic in practice. + +" +Where is the explanation about Figure 1(f)? What is p_init? +" +This is a typo. It should read "n" the number of cascades. + +" +The authors need to show at least one common metric for all types for graphs +" +This is an interesting point: A graph plotting the same metric for all +considered networks can replace one of the 6 figures. + +Misc. +" +Citations/Related work remarks +" +The requested citations can be included on lines 42, 68, 75, 78, 93, 362. The +authors regret not to have cited Du et al. 2012 and their work should be +included in the related work section along with other work considering the +estimation of influence in networks. It can be mentioned that Daneshmand et al +adopt the same model as Gomez-R et al '10 and Abrahao et al. '13. The phrasing +can be changed from "Graph Inference" to "Network Inference" with the requested +citations. |