# Correction to "Gaussian Process Dynamical Models for Human Motion"

Jack M.

David J. , IEEE

Aaron , IEEE

Pages: p. 1118

Due to a publication error in [ ^{1}], two figures were misprinted. Figs. 7 and 8 on p. 292 should read shown below:

Figure Fig. 7. Models learned with fixed $\bar \alpha$ from three different walking subjects. (a) The learned latent coordinates shown in blue. (b) $-\ln$ variance plot shows smooth high confidence regions, but the variance near data is larger than in Fig. 5c, similar to B-GPDM. (c) Typical samples from the dynamic predictive distribution are shown in green, while the mean-prediction sample is shown in red.

Figure Fig. 8. Models learned with two-stage MAP from four different walking subjects. (a) The learned latent coordinates shown in blue, note the walkers are separated into distinct portions of the latent space. (b) $-\ln$ variance plot shows smooth high confidence regions, and the variance near data is similar to Fig. 5c. (c) Typical samples from the dynamic predictive distribution are shown in green, while the mean-prediction sample is shown in red.

## REFERENCES

- 1. J.M. Wang, David J. Fleet, and A. Hertzmann, "Gaussian Process Dynamical Models for Human Motion,"
*IEEE Trans. Pattern Analysis and Machine Intelligence,* vol. 30, no. 2, pp. 283-298, Feb. 2008.