I'd also suggest to mark the distributed ALS code as experimental, because
we have not yeen seen good results yet, probably because of the iterative
nature of the algorithm.
Am 17.05.2011 21:37 schrieb "Dmitriy Lyubimov" <[email protected]>:
> Our company just had a new hire whose tasks would include a somewhat
> sizeable inputs on SSVDs so hopefully we'll get more data soon and
> validation soon.
>
> On Tue, May 17, 2011 at 12:34 PM, Dmitriy Lyubimov <[email protected]>
wrote:
>> We still haven't got a good size test confirmation for SSVD (at least
>> Mahout's version of it) so i'd put experimental and expect hotfixes.
>>
>> On Tue, May 17, 2011 at 7:50 AM, Sean Owen <[email protected]> wrote:
>>> • Improved Lanczos solver
>>> • Stochastic Singular Value Decomposition implementation
>>> • Incremental SVD implementation
>>> • Alternating Least Squares with Weighted Regularization
>>> collaborative filtering implementation, both distributed and
>>> non-distributed
>>> • SVDRecommender enhancements
>>> • Better control over candidate item selection in
item-based recommenders
>>> • Significant removal of deprecated or dead code
>>> • Many bug fixes, refactorings and other small improvements
>>>
>>> What else?
>>>
>>
