Tag
This paper proposes a data-driven elastic-net support vector machine that learns simplex-constrained weights over candidate pinball losses, with a distributed solver for vertically partitioned high-dimensional data. Theoretical guarantees and experiments demonstrate equivalence to centralized training under common initialization.
This paper identifies feature starvation in sparse autoencoders as a geometric instability and proposes adaptive elastic net SAEs (AEN-SAEs) to mitigate it without heuristics.