MeganckSajdakWu - Revision history

imported>Projects221: /* Appendix I: Source Code */

2014-03-17T05:00:40Z

Appendix I: Source Code

← Older revision		Revision as of 05:00, 17 March 2014
Line 226:		Line 226:
	*LIBSVM		*LIBSVM

	The presentation may be ~~downloaded as a [[Media:Megancksajdakwu.pptx\|.pptx]] or as a~~ [[Media:Megancksajdakwu.pdf\|~~.pdf~~]].		The presentation may be [[Media:Megancksajdakwu.pdf\|downloaded here]].

	== Appendix II: Work Breakdown ==		== Appendix II: Work Breakdown ==

imported>Projects221: /* Appendix I: Source Code */

2014-03-17T04:58:34Z

Appendix I: Source Code

← Older revision		Revision as of 04:58, 17 March 2014
Line 225:		Line 225:
	*ISETBIO		*ISETBIO
	*LIBSVM		*LIBSVM

			The presentation may be downloaded as a [[Media:Megancksajdakwu.pptx\|.pptx]] or as a [[Media:Megancksajdakwu.pdf\|.pdf]].

	== Appendix II: Work Breakdown ==		== Appendix II: Work Breakdown ==

imported>Projects221: /* Conclusions */

2014-03-17T04:57:30Z

Conclusions

← Older revision		Revision as of 04:57, 17 March 2014
Line 186:		Line 186:

	*Results match the current trends in the markets, in that monitors have about half the ppi of tablets, and TV's have half the ppi of monitors. However, in all cases current markets have below the threshold PPI, and therefore have not met their full potential.		*Results match the current trends in the markets, in that monitors have about half the ppi of tablets, and TV's have half the ppi of monitors. However, in all cases current markets have below the threshold PPI, and therefore have not met their full potential.
	*Simulations agree with published results		*Simulations agree with published results.
	*Support Vector Machine and Nearest Neighbor have similar performance in the transition zone		*Support Vector Machine and Nearest Neighbor have similar performance in the transition zone.
	*The effect of contrast on Vernier Acuity appears to be nonlinear and matches well with qualitative results		*The effect of contrast on Vernier Acuity appears to be nonlinear and matches well with qualitative results.
	*Pupil diameter seems to have a marginal effect on performance, which implies that acuity doesn't really change with different levels of ambient light		*Pupil diameter seems to have a marginal effect on performance, which implies that acuity doesn't really change with different levels of ambient light.

	=== Further Investigation ===		=== Further Investigation ===

imported>Projects221: /* Conclusions */

2014-03-17T04:54:06Z

Conclusions

← Older revision		Revision as of 04:54, 17 March 2014
Line 183:		Line 183:
	== Conclusions ==		== Conclusions ==

	~~Both~~ machine learning algorithms ~~displayed~~ comparable performance with respect to the Vernier acuity task, and the results are similar to those of previous published experiments. It is important to note that the resolution threshold for Vernier acuity is strongly dependent on (in fact, nearly proportional to) the distance between the observer and screen.		The two machine learning algorithms were very comparable in performance with respect to the Vernier acuity task, and the results are similar to those of previous published experiments. It is important to note that the resolution threshold for Vernier acuity is strongly dependent on (in fact, nearly proportional to) the distance between the observer and screen.

	*results match the current trends in the desktop/tv/tablet/phone market. talk about 4K and increasing phone display resolution.
	*both algorithms have similar performance in the transition zone
	*results agree with published results
	*contrast appears to be nonlinear and matches well with qualitative results
	*inconclusive with pupil diameter, which implies that acuity doesn't really change with different levels of light

	~~MOAR~~

			*Results match the current trends in the markets, in that monitors have about half the ppi of tablets, and TV's have half the ppi of monitors. However, in all cases current markets have below the threshold PPI, and therefore have not met their full potential.
			*Simulations agree with published results
			*Support Vector Machine and Nearest Neighbor have similar performance in the transition zone
			*The effect of contrast on Vernier Acuity appears to be nonlinear and matches well with qualitative results
			*Pupil diameter seems to have a marginal effect on performance, which implies that acuity doesn't really change with different levels of ambient light

	=== Further Investigation ===		=== Further Investigation ===

imported>Projects221: /* Conclusions */

2014-03-16T23:46:28Z

Conclusions

← Older revision		Revision as of 23:46, 16 March 2014
Line 189:		Line 189:
	*results agree with published results		*results agree with published results
	*contrast appears to be nonlinear and matches well with qualitative results		*contrast appears to be nonlinear and matches well with qualitative results
			*inconclusive with pupil diameter, which implies that acuity doesn't really change with different levels of light

	MOAR		MOAR

imported>Projects221: /* Conclusions */

2014-03-16T23:45:40Z

Conclusions

← Older revision		Revision as of 23:45, 16 March 2014
Line 185:		Line 185:
	Both machine learning algorithms displayed comparable performance with respect to the Vernier acuity task, and the results are similar to those of previous published experiments. It is important to note that the resolution threshold for Vernier acuity is strongly dependent on (in fact, nearly proportional to) the distance between the observer and screen.		Both machine learning algorithms displayed comparable performance with respect to the Vernier acuity task, and the results are similar to those of previous published experiments. It is important to note that the resolution threshold for Vernier acuity is strongly dependent on (in fact, nearly proportional to) the distance between the observer and screen.

	results match the current trends in the desktop/tv/tablet/phone market. talk about 4K and increasing phone display resolution.		*results match the current trends in the desktop/tv/tablet/phone market. talk about 4K and increasing phone display resolution.
	both algorithms have similar performance in the transition zone		*both algorithms have similar performance in the transition zone
	results agree with published results		*results agree with published results
			*contrast appears to be nonlinear and matches well with qualitative results

	MOAR		MOAR

imported>Projects221: /* Conclusions */

2014-03-16T23:44:05Z

Conclusions

← Older revision		Revision as of 23:44, 16 March 2014
Line 185:		Line 185:
	Both machine learning algorithms displayed comparable performance with respect to the Vernier acuity task, and the results are similar to those of previous published experiments. It is important to note that the resolution threshold for Vernier acuity is strongly dependent on (in fact, nearly proportional to) the distance between the observer and screen.		Both machine learning algorithms displayed comparable performance with respect to the Vernier acuity task, and the results are similar to those of previous published experiments. It is important to note that the resolution threshold for Vernier acuity is strongly dependent on (in fact, nearly proportional to) the distance between the observer and screen.

	~~Compare to~~ current ~~tablets~~/~~desktops~~/~~TVs~~		results match the current trends in the desktop/tv/tablet/phone market. talk about 4K and increasing phone display resolution.
			both algorithms have similar performance in the transition zone
			results agree with published results

	MOAR		MOAR

imported>Projects221: /* Further Investigation */

2014-03-16T23:36:30Z

Further Investigation

← Older revision		Revision as of 23:36, 16 March 2014
Line 198:		Line 198:

	;Neural network and other algorithms		;Neural network and other algorithms
	:Only two machine learning algorithms were tested in this experiment. Implementation a neural network (i.e. with backpropagation) may yield interesting and useful results.		:Only two machine learning algorithms were tested in this experiment. Implementation of a neural network (i.e. with backpropagation) may yield interesting and useful results.

	;Modeling myopia and hyperopia		;Modeling myopia and hyperopia

imported>Projects221: /* Appendix II: Work Breakdown */

2014-03-14T04:53:03Z

Appendix II: Work Breakdown

← Older revision		Revision as of 04:53, 14 March 2014
Line 243:		Line 243:
	* ISET / ISETBIO investigation		* ISET / ISETBIO investigation
	* Training / testing		* Training / testing
	* Data analysis		* Data analysis / interpretation

imported>Projects221: /* Appendix I: Source Code */

2014-03-14T04:45:41Z

Appendix I: Source Code

← Older revision		Revision as of 04:45, 14 March 2014
Line 218:		Line 218:

	== Appendix I: Source Code ==		== Appendix I: Source Code ==
	Source code can be [[Media:Vernier-modeling.zip\|downloaded here]] or ~~cloned~~ from https://github.com/shwu/vernier-modeling.git		Source code can be [[Media:Vernier-modeling.zip\|downloaded here]] or obtained from GitHub at https://github.com/shwu/vernier-modeling.git

	The following MATLAB libraries are required to run the code:		The following MATLAB libraries are required to run the code: