![](bg8.png)
Table 5: T-test applied to the results between methods BO and MCNN-GA.
Accuracy Loss Size
Inference
Time
Execution
Time
Fitness
t-value p-value t-value p-value t-value p-value t-value p-value t-value p-value t-value p-value
-3.125 0.006 3.048 0.007 4.455 0.0003 0.0 1.0 -3.966 0.001 5.596 0.00003
Table 6: MCNN-GA and ThiNet30 comparison results
(mean value).
Method Accuracy Loss Size Inference Time
MCNN-GA 99.35 0.099 35026.0 5.4
ThiNet30 99.10 0.113 6700576 6.5
Table 7: T-test applied to the results between methods
MCNN-GA and ThiNet30.
Accuracy Loss Size
Inference
Time
t-value p-value t-value p-value t-value p-value t-value p-value
1.983 0.062 -1.199 0.245 -9968.12 0 -5.312 0.00004
networks. The case study used to validate the analysis
was the Dota2 game event dataset. The results showed
that MCNN-GA generated CNNs which achieved a
classification performance as good as the best model
produced in (Luo et al., 2019a) (ThiNet30), but with
significantly fewer parameters (resulting in models
with less memory usage). It means that MCNN-GA
have a great potential to generate highly efficient and
suitable models for real-time applications (which can
be transferable to domains beyond games), which
helps in hardware accessibility for complex tasks.
As future works, the authors intend: to investi-
gate the performance of the approaches studied here
for different types of games (simpler 2D games like
Super Mario Bros and more realistic 3D games like
Skyrim); to publish the framework investigated here
for the community’s use; and, finally, to use such ap-
proaches to build real-time mechanisms that can help
people with cognitive difficulties.
REFERENCES
Azadvar, A. (2021). Predictive psychological player profil-
ing.
Coleman, A. M. (2009). A dictionary of psychology / An-
drew M. Colman. Oxford University Press Oxford ;
New York, 3rd ed. edition.
Deng, J., Dong, W., Socher, R., Li, L.-J., Li, K., and Fei-Fei,
L. (2009). Imagenet: A large-scale hierarchical image
database. In 2009 IEEE conference on computer vi-
sion and pattern recognition, pages 248–255.
Faria, M. P. P., Julia, R. M. S., and Tomaz, L. B. P.
(2020). Improving fifa player agents decision-making
architectures based on convolutional neural networks
through evolutionary techniques. In Cerri, R. and
Prati, R. C., editors, Intelligent Systems, pages 371–
386, Cham. Springer International Publishing.
Han, S., Mao, H., and Dally, W. J. (2016). Deep compres-
sion: Compressing deep neural network with pruning,
trained quantization and huffman coding. In 4th Inter-
national Conference on Learning Representations.
He, K., Zhang, X., Ren, S., and Sun, J. (2016). Deep resid-
ual learning for image recognition. In Proceedings of
the IEEE Conference on Computer Vision and Pattern
Recognition (CVPR).
Howard, A. G., Zhu, M., Chen, B., Kalenichenko, D.,
Wang, W., Weyand, T., Andreetto, M., and Adam,
H. (2017). Mobilenets: Efficient convolutional neu-
ral networks for mobile vision applications.
Luo, J.-H., Wu, J., and Lin, W. (2017). Thinet: A filter level
pruning method for deep neural network compression.
In 2017 IEEE International Conference on Computer
Vision (ICCV), pages 5068–5076.
Luo, Z., Guzdial, M., Liao, N., and Riedl, M. (2018). Player
experience extraction from gameplay video. CoRR,
abs/1809.06201.
Luo, Z., Guzdial, M., and Riedl, M. (2019a). Making cnns
for video parsing accessible. CoRR, abs/1906.11877.
Luo, Z., Guzdial, M., and Riedl, M. (2019b). Making
cnns for video parsing accessible: Event extraction
from dota2 gameplay video using transfer, zero-shot,
and network pruning. In Proceedings of the 14th In-
ternational Conference on the Foundations of Digital
Games, FDG ’19, New York, NY, USA. Association
for Computing Machinery.
Miller, B. L., Miller, B. L., Goldberg, D. E., and Gold-
berg, D. E. (1995). Genetic algorithms, tournament
selection, and the effects of noise. Complex Systems,
9:193–212.
Mockus, J. B. and Mockus, L. J. (1991). Bayesian ap-
proach to global optimization and application to mul-
tiobjective and constrained problems. J. Optim. The-
ory Appl., 70(1):157–172.
Moosa, A. M., Al-Maadeed, N., Saleh, M., Al-Maadeed,
S. A., and Aljaam, J. M. (2020). Designing a mobile
serious game for raising awareness of diabetic chil-
dren. IEEE Access, 8:222876–222889.
Soomro, K., Zamir, A., and Shah, M. (2012). Ucf101: A
dataset of 101 human actions classes from videos in
the wild. CoRR.
Srinivas, M. and Patnaik, L. M. (1994). Genetic algorithms:
a survey. Computer, 27:17–26.
Wijman, T. (2019). The global games market will generate
$152.1 billion in 2019 as the u.s. overtakes china as
the biggest market.
Wong, J. and Gales, M. (2016). Sequence student-teacher
training of deep neural networks. pages 2761–2765.
Xu, M., Gao, M., Chen, Y.-T., Davis, L. S., and Crandall,
D. J. (2019). Temporal recurrent networks for online
action detection.
VISAPP 2023 - 18th International Conference on Computer Vision Theory and Applications
888