Not today. However, there are a few things that can be done to improve speed or performance if one does not want to use Trifacta's advanced machine-learning algorithms to suggest potential join methods.
One example would be to point to a small, empty datasets, or parameterized datasets; then manually define your join; and then subsequently point back to the actual datasets.
Another would be to use Trifacta's advanced sampling capabilities to reduce the size of previewed data. You can always resample after the join is performed to continue iterating and wrangling fast--this would be a sample taken after the full datasets are joined.
