Tomer Kaftan

After talking to @etrain and @shivaram it turns out I was mistaken about the interface of the BlockLinearEstimator. So, concatenate should actually have the interface: `Seq[(T => DenseVector)] => (T...

Okay @shivaram and @etrain, as per our discussion I’ve gone through MnistRandomFFT, RandomPatchCifar, VOCSIFTFisher, and the Imagenet pipeline to figure out how to better linearize our existing pipelines. ### Lessons...

@etrain and @shivaram How would you guys feel about this: ``` scala def reducePoint[T : FloatOrDouble](x: DenseVector[T], mat: DenseMatrix[T]): DenseVector[T] = { val ops = implicitly[FloatOrDouble[T]] import ops._ (x.t *...

I'm not sure. It's a little unclear to me how scala implicit imports work, and it looks like we have to do the implicitly import within the method — Best...

I'm able to get this to work: ``` scala case class LinearMapper[T: BreezeNumeric](x: DenseMatrix[T]) extends NumericTransformer[T] { import ops._ /** * Apply a linear model to an input. * @param...

Hmm with an extra ~100 lines of code in NumericTransformer individually referencing every implicit def in MutableOptimizationSpace I can make it so the subclasses don't need `import ops._`. Thoughts?

Until breeze changes the interface of MutableOptimizationSpace. (I have no idea how likely that is though)

The problem is actually somewhat unrelated and comes down to something along the lines of the following: - to capture multiple implicits via a single context bound (or a single...

Okay, I've run into some issues w/ broadcasting, but hopefully will get a response soon to my Breeze issue.

Side note: It concerns me how much I've learned about scala implicits -_-