[Predator] Making the Y sets conditional for loglinear models

[Predator] Making the Y sets conditional for loglinear models

Because we adjusted things so that features return -inf whenever they want
to ensure that a given Y/Suspect is never blamed, we can improve things
so that we only need to iterate over the subset of Y/Suspect which does
not have -inf scores (i.e., does not have zero probability). This CL makes
that change: the (normalized) LogLinearModel and TrainableLogLinearModel
now take a function X -> iterable(Y) where all y with non-zero probability
are in the iterable; as opposed to the old iterable(Y) that enumerates
all possible Y.

BUG=
TBR=stgao@chromium.org

Review-Url: https://codereview.chromium.org/2595263003
Committed: https://chromium.googlesource.com/infra/infra/+/49e6c276ef9428af115e0cbd7e77e9a4d2b0d633

[wrengr, 2016-12-22 20:24:01]

Description was changed from

==========
[Predator] Making the Y sets conditional for loglinear models

Because we adjusted things so that features return -inf whenever they want
to ensure that a given Y/Suspect is never blamed, we can improve things
so that we only need to iterate over the subset of Y/Suspect which does
not have -inf scores (i.e., does not have zero probability). This CL makes
that change: the (normalized) LogLinearModel and TrainableLogLinearModel
now take a function X -> iterable(Y) where all y with non-zero probability
are in the iterable; as opposed to the old iterable(Y) that enumerates
all possible Y.

BUG=
TBR=stgao@chromium.org
==========

to

==========
[Predator] Making the Y sets conditional for loglinear models

Because we adjusted things so that features return -inf whenever they want
to ensure that a given Y/Suspect is never blamed, we can improve things
so that we only need to iterate over the subset of Y/Suspect which does
not have -inf scores (i.e., does not have zero probability). This CL makes
that change: the (normalized) LogLinearModel and TrainableLogLinearModel
now take a function X -> iterable(Y) where all y with non-zero probability
are in the iterable; as opposed to the old iterable(Y) that enumerates
all possible Y.

BUG=
TBR=stgao@chromium.org
==========

[wrengr, 2016-12-22 20:24:01]

wrengr@chromium.org changed reviewers:
+ inferno@chromium.org, katesonia@chromium.org, mbarbella@chromium.org

[wrengr, 2016-12-22 20:24:12]

PTAL

[Sharu Jiang, 2016-12-23 00:06:21]

lgtm

[wrengr, 2016-12-27 17:24:49]

The CQ bit was checked by wrengr@chromium.org

[commit-bot: I haz the power, 2016-12-27 17:24:59]

CQ is trying da patch. Follow status at
 
https://chromium-cq-status.appspot.com/v2/patch-status/codereview.chromium.or...

[commit-bot: I haz the power, 2016-12-27 17:34:03]

CQ is committing da patch.

Bot data: {"patchset_id": 20001, "attempt_start_ts": 1482859489880020,
"parent_rev": "d5816b11ae474053b33f3a69a5b864e1ceb92a53", "commit_rev":
"49e6c276ef9428af115e0cbd7e77e9a4d2b0d633"}

[commit-bot: I haz the power, 2016-12-27 17:34:05]

Description was changed from

==========
[Predator] Making the Y sets conditional for loglinear models

Because we adjusted things so that features return -inf whenever they want
to ensure that a given Y/Suspect is never blamed, we can improve things
so that we only need to iterate over the subset of Y/Suspect which does
not have -inf scores (i.e., does not have zero probability). This CL makes
that change: the (normalized) LogLinearModel and TrainableLogLinearModel
now take a function X -> iterable(Y) where all y with non-zero probability
are in the iterable; as opposed to the old iterable(Y) that enumerates
all possible Y.

BUG=
TBR=stgao@chromium.org
==========

to

==========
[Predator] Making the Y sets conditional for loglinear models

Because we adjusted things so that features return -inf whenever they want
to ensure that a given Y/Suspect is never blamed, we can improve things
so that we only need to iterate over the subset of Y/Suspect which does
not have -inf scores (i.e., does not have zero probability). This CL makes
that change: the (normalized) LogLinearModel and TrainableLogLinearModel
now take a function X -> iterable(Y) where all y with non-zero probability
are in the iterable; as opposed to the old iterable(Y) that enumerates
all possible Y.

BUG=
TBR=stgao@chromium.org

Review-Url: https://codereview.chromium.org/2595263003
Committed:
https://chromium.googlesource.com/infra/infra/+/49e6c276ef9428af115e0cbd7e77e...
==========

[commit-bot: I haz the power, 2016-12-27 17:34:06]

Committed patchset #2 (id:20001) as
https://chromium.googlesource.com/infra/infra/+/49e6c276ef9428af115e0cbd7e77e...