Filename ▼	Summary + Labels ▼	Uploaded ▼	ReleaseDate ▼	Size ▼	DownloadCount ▼	...
	sa_final_feature_selection_vs_expected_accuracy_for_different_dataset.png	"orthogonal" feature selection - final feature selection (min cuis in a doc) vs expected accuracy for different dataset   NLP DM ML FS	Jan 24	Jan 24	101 KB	6
	not_much_improvement_above_3000.png	Not much improvement above 3000 training positive and negative samples   NLP ML INSTANCE SELECTION	Jan 12	Jan 12	443 KB	5
	all_77_class_centers_11990_feats.png	All 77 HD class centers MDS (cosine of class CUI sums dist) comparison using 11990 features   NLP ML	Jan 5	Jan 5	144 KB	5
	all_77_class_centers_14339_feats.png	All 77 HD class centers MDS (cosine of class CUI sums dist) comparison using 14339 features   NLP ML	Jan 5	Jan 5	144 KB	5
	data_size_vs_nsv.png	Artificial data - two gauss data, dependency of size and number of support vectors and margin cleanup   ML DM SVM	Oct 2011	Oct 2011	70.4 KB	6
	forward_selection_issues.png	Issues with estimation when adding new features in CV loop.   ML FS TC	Oct 2011	Oct 2011	26.3 KB	5
	simon_effect_large_p.png	So called Simon (2003) effect - Large p dependency	Sep 2011	Sep 2011	25.6 KB	6
	simon_effect_sparse.png	So colled Simon (2003) effect - sparcisity dependence	Sep 2011	Sep 2011	31.6 KB	5
	simple_simon.png	So called Simon (2003) effect - F1 dependence	Sep 2011	Sep 2011	35.4 KB	5
	wrapper_improvement_diff_thrsh_add.png	Different improvement wrapper feature selection rates based on S.A. thresholding techniques - almost additive nature of improvement   NLP ML SA	Sep 2011	Sep 2011	57.0 KB	5
	wrapper_improvement_diff_thrsh.png	Different improvement wrapper feature selection rates based on S.A. thresholding techniques   ML NLP SA	Sep 2011	Sep 2011	36.6 KB	5
	near_additive_nature_of_spreading_activation.png	Near additive nature of spreading activation (works only with SVM wrapper feats. sel.)   NLP ML FS	Sep 2011	Sep 2011	78.9 KB	5
	correlation_between_R600_and_R900_wrapper.png	Relation between wrapper feature selection improvement after 600 and 900 samples   NL NLP FS	Sep 2011	Sep 2011	31.3 KB	9
	wrapper_vs_filter_correlation.png	Relation between wrapper feature selection improvement and simple filter (all of them look as bad)   ML NLP FS	Sep 2011	Sep 2011	27.1 KB	11
	spr_act_has_additive_wrapp_improv.jpg	Spreading activation has near additive improvement properties - two good feature give - 2x improvement   NLP ML DM SA	Aug 2011	Aug 2011	36.5 KB	7
	small_layered_memory_map_for_hd_hard.png	Layered version (spreading activation steps) of neruocognitive pathways for two class problem ("endocarditis,_bacterial" "heart_valve_diseases") - trimmed by frequency   NLP ML TM SA	Aug 2011	Aug 2011	425 KB	8
	small_memory_map_for_hd_hard.png	Smaller version of neruocognitive pathways for two class problem ("endocarditis,_bacterial" "heart_valve_diseases") - trimmed by frequency   NLP ML SA TM	Aug 2011	Aug 2011	218 KB	9
	big_memory_map_for_hd_hard.png	Big version of neruocognitive pathways for two class problem ("endocarditis,_bacterial" vs "heart_valve_diseases")   NLP ML SA	Aug 2011	Aug 2011	556 KB	5
	dlfs_artif_data_small.jpg	Local feature selection for sparse data that simulates classification data with two classes and multiple sub-topics   NLP DM Selection	Jun 2011	Jun 2011	610 KB	6
	real_data_hd_medium.jpg	Direct local feature selection on real data (pubmed, ohsumed, hd) compared with information gain   NLP ML TM	Jun 2011	Jun 2011	989 KB	7
	feats_reduction_direct_local_fs.jpg	Direct local features selection on different size subtopics   NLP Learning DM	Jun 2011	Jun 2011	440 KB	7
	zzimg.pca_tr_te_scatter.z.gif	Spreading activation - feature space projections   NLP ML SA	Jun 2011	Jun 2011	135 KB	10
	label_co_occurance_pubmed.png	Label co-occurance in OHSUMED Heart Diseases sub-set.   NLP ML DM	Jun 2011	Jun 2011	188 KB	7