Command-Line Parameter Index

Each listing displays the parameter's name, syntax, a description of acceptable arguments, its default setting (if any), and a page reference to a more detailed description in Chapter 5.

Name	Syntax	Argument Description	Default	Page
Architecture file	-A <file>	Input filename.	none
Also write pending features	-a $\pm$	`+' or `-'.	-
Naive Bayes	-B ...	See Chapter 5.	none	,
Naive Bayes smoothing	-b <k>	A real number.	15
Learning curve generation	-c <i>	A non-negative integer.	0
Discard method	-d ...	See Chapter 5.	none
Error file	-E <file>	Output filename.	none
Eligibility policy	-e ...	See Chapter 5.	count:2
``Fixed'' feature	-f $\pm$	`+' or `-'.	+
Gradient Descent	-G $\pm$	`+' or `-'.	-	,
Generate conjunctions	-g $\pm[,\pm]$	`+' or `-' once or twice.	<unset>,-
Incremental learning	-i $\pm$	`+' or `-'.	-
Target output limit	-L <k>	A positive integer.	ULONG_MAX
Labeled examples	-l $\pm$	`+' or `-'.	+
Store examples in memory	-M $\pm$	`+' or `-'.	-
Multiple labels	-m $\pm$	`+' or `-'.	+
Constraint Classification	-O $\pm[,\pm]$	`+' or `-' once or twice.	-,-	,
Output mode	-o ...	See Chapter 5.	accuracy
Perceptron	-P ...	See Chapter 5.	none	,
Prediction threshold	-p <k>	A real number.	$-1$
Results file	-R <file>	Output filename.	none
Training rounds (cycles)	-r <i>	A positive integer.	$2$
Thick separator	-S <p>$[$,<n>$]$	One or two real numbers.	0	,
Sparse network	-s <s | f>	`s' or `f'.	s
Testing file	-T <file>	Input filename.	none
Threshold relative updating	-t $\pm$	`+' or `-'.	-	,
First cycle updating	-u $\pm$	`+' or `-'.	+
Verbosity	-v ...	See Chapter 5.	min
Winnow	-W ...	See Chapter 5.	none	,
Winnow smoothing	-w <k>	A real number.	0
``Raw'' mode	-z $\pm$	`+' or `-'.	-