updates

Author: jcanny

src/main/scala/BIDMach/models/GLM.scala

@@ -44,18 +44,18 @@ class GLM(opts:GLM.Opts) extends RegressionModel(opts) {
   def mupdate(in:Mat) = {
     val targs = targets * in
     min(targs, 1f, targs)
-    val alltargs = if (targmap.asInstanceOf[AnyRef] != null) targmap * targs else targs
     val dweights = if (iweight.asInstanceOf[AnyRef] != null) iweight * in else null
-    mupdate3(in, alltargs, dweights)
+    mupdate3(in, targs, dweights)
   }
 
   def mupdate2(in:Mat, targ:Mat) = mupdate3(in, targ, null)
 
-  def mupdate3(in:Mat, targ:Mat, dweights:Mat) = {
-    val ftarg = full(targ)
+  def mupdate3(in:Mat, targ:Mat, dweights:Mat) = {        
+    val ftarg = full(targ);
+    val targs = if (targmap.asInstanceOf[AnyRef] != null) targmap * ftarg else ftarg;
     val eta = modelmats(0) * in  
     GLM.preds(eta, eta, mylinks, linkArray, totflops)
-    GLM.derivs(eta, ftarg, eta, mylinks, linkArray, totflops)
+    GLM.derivs(eta, targs, eta, mylinks, linkArray, totflops)
     if (dweights.asInstanceOf[AnyRef] != null) eta ~ eta ∘  dweights
     updatemats(0) ~ eta *^ in
     if (mask.asInstanceOf[AnyRef] != null) {
@@ -64,21 +64,21 @@ class GLM(opts:GLM.Opts) extends RegressionModel(opts) {
   }
 
   def meval(in:Mat):FMat = {
-    val targs = targets * in
-    min(targs, 1f, targs)
-    val alltargs = if (targmap.asInstanceOf[AnyRef] != null) targmap * targs else targs
-    val dweights = if (iweight.asInstanceOf[AnyRef] != null) iweight * in else null
-    meval3(in, alltargs, dweights)
+    val targs = targets * in;
+    min(targs, 1f, targs);
+    val dweights = if (iweight.asInstanceOf[AnyRef] != null) iweight * in else null;
+    meval3(in, targs, dweights);
   }
 
   def meval2(in:Mat, targ:Mat):FMat = meval3(in, targ, null)
 
   def meval3(in:Mat, targ:Mat, dweights:Mat):FMat = {
-    val ftarg = full(targ)
+    val ftarg = full(targ);
+    val targs = if (!(putBack >= 0) && targmap.asInstanceOf[AnyRef] != null) targmap * ftarg else ftarg;
     val eta = modelmats(0) * in
     GLM.preds(eta, eta, mylinks, linkArray, totflops)
-    val v = GLM.llfun(eta, ftarg, mylinks, linkArray, totflops)
-    if (putBack >= 0) {ftarg <-- eta}
+    val v = GLM.llfun(eta, targs, mylinks, linkArray, totflops)
+    if (putBack >= 0) {targ <-- eta}
     if (dweights.asInstanceOf[AnyRef] != null) {
       FMat(sum(v ∘  dweights, 2) / sum(dweights))
     } else {

src/main/scala/BIDMach/models/Regression.scala

@@ -27,7 +27,13 @@ abstract class RegressionModel(override val opts:RegressionModel.Opts) extends M
     val data0 = mats(0)
     val m = size(data0, 1)
     val targetData = mats.length > 1
-    val d = if (targetData) mats(1).nrows else if (opts.targmap.asInstanceOf[AnyRef] != null) opts.targmap.nrows else opts.targets.nrows 
+    val d = if (opts.targmap.asInstanceOf[AnyRef] != null) {
+      opts.targmap.nrows 
+    } else if (opts.targets.asInstanceOf[AnyRef] != null) {
+      opts.targets.nrows 
+    } else {
+      mats(1).nrows  
+    }
     val sdat = (sum(data0,2).t + 0.5f).asInstanceOf[FMat]
     sp = sdat / sum(sdat)
     println("corpus perplexity=%f" format (math.exp(-(sp ddot ln(sp)))))
@@ -39,9 +45,9 @@ abstract class RegressionModel(override val opts:RegressionModel.Opts) extends M
     }
     updatemats = new Array[Mat](1)
     updatemats(0) = modelmats(0).zeros(modelmats(0).nrows, modelmats(0).ncols)
+    targmap = if (useGPU && opts.targmap.asInstanceOf[AnyRef] != null) GMat(opts.targmap) else opts.targmap
     if (! targetData) {
       targets = if (useGPU && opts.targets.asInstanceOf[AnyRef] != null) GMat(opts.targets) else opts.targets
-      targmap = if (useGPU && opts.targmap.asInstanceOf[AnyRef] != null) GMat(opts.targmap) else opts.targmap
       mask =    if (useGPU && opts.rmask.asInstanceOf[AnyRef] != null) GMat(opts.rmask) else opts.rmask
     }
   }