Author Affiliations
1School of Internet of Things, Jiangnan University, Wuxi, Jiangsu 214122, China2Engineering Research Center of Internet of Things Technology Applications of Ministry of Education, Wuxi, Jiangsu 214122, Chinashow less
Fig. 1. Overall framework of gesture identity authentication
Fig. 2. Different types of CDBN models structure. (a) General structure; (b) with pooling layer
Fig. 3. Different model structures. (a) RBM structure; (b) CRBM structure
Fig. 4. Maximum pooling operation
Fig. 5. Single step Gibbs sampling
Fig. 6. Structure diagram of the output authentication model
Fig. 7. User 1 original motion trajectory
Fig. 8. User 1 pre-processed trajectory
Fig. 9. Fake user's gesture recovery diagram
Depth | ACC /% | FAR /% | FRR /% | Time /s |
---|
1 | 93.524 | 6.22 | 8.00 | 60.90 | 2 | 97.667 | 2.22 | 3.33 | 130.56 | 3 | 92.333 | 7.11 | 10.67 | 251.12 |
|
Table 1. Simulation results of different network depths
Index | ACC /% | FAR /% | FRR /% | Time /s |
---|
0.010 | 96.333 | 2.80 | 8.00 | 109.36 | 0.015 | 97.000 | 2.00 | 8.00 | 115.98 | 0.020 | 97.667 | 2.22 | 3.33 | 130.56 | 0.025 | 95.000 | 0 | 30.00 | 134.02 | 0.200 | 83.333 | 0 | 100.00 | 139.88 |
|
Table 2. Simulation results of different sparsity indices
Method | ACC /% | FAR /% | FRR /% | Time /s |
---|
None | 97.333 | 2.00 | 6.00 | 400.31 | Mean | 97.652 | 2.33 | 3.35 | 130.77 | Max | 97.667 | 2.22 | 3.33 | 130.56 |
|
Table 3. Simulation results of different pooling methods
Epoch | ACC /% | FAR /% | FRR /% | Time /s |
---|
5∈ | 96.000 | 0.80 | 20.00 | 70.33 | 10∈ | 97.667 | 2.22 | 3.33 | 130.56 | 25∈ | 96.000 | 2.80 | 10.00 | 375.61 | 50∈ | 97.000 | 2.00 | 8.00 | 643.80 |
|
Table 4. Simulation results of different iteration times
Layer | ACC /% | FAR /% | FRR /% | Time /s |
---|
Fully | 97.333 | 2.67 | 2.67 | 367.44 | RMS | 97.667 | 2.22 | 3.33 | 130.56 |
|
Table 5. Simulation results of different connection layers
Method | ACC /% | FAR /% | FRR /% | Time /s |
---|
BP | 92.660 | 4.53 | 5.01 | 56.84 | HMM | 93.250 | 3.44 | 4.46 | 85.69 | DBN | 96.630 | 2.37 | 3.79 | 119.31 | CDBN | 97.667 | 2.22 | 3.33 | 130.56 |
|
Table 6. Performance comparison among CDBN algorithm, BP, HMM, and DBN algorithms