A locally-built, LLM-digested index of recent arXiv papers in quant finance, geometry/topology, and statistical ML — keyword search served straight from SQLite on this machine.
A new mechanism for GANs improves text generation by evaluating sub-sequences.
GGP models multivariate time series with latent sub-sequences for diverse behaviors.
In \cite{CM5}, Colding and Minicozzi describe a type of compactness property possessed by sequences of embedded minimal surfaces in $\Real^3$ with finite genus and with boundaries going to . They show that any such sequence either contains a sub-sequence with uniformly bounded curvature or the sub-sequence has …
A framework to explain decoder-only sequence classification models using intermediate predictions.
This paper describes the behavior of sequences of solutions to the Kapustin-Witten equations with Nahm pole asymptotics on the product of the half-line with a compact, oriented, Riemannian 3-manifold. These sequences have sub-sequences that either converge to another solution after acting term-wise by an automorphism o…
SAttention improves long sequence attention with smoothed skeleton sketching.
This paper explores and ties together three themes. The first is to establish regularity of a metric tensor, on a manifold with boundary, on which there are given Ricci curvature bounds, on the manifold and its boundary, and a Lipschitz bound on the mean curvature of the boundary. The second is to establish geometric c…
Many machine learning tasks require sampling a subset of items from a collection based on a parameterized distribution. The Gumbel-softmax trick can be used to sample a single item, and allows for low-variance reparameterized gradients with respect to the parameters of the underlying distribution. However, stochastic o…
Interleaved RNNs detect fraud without costly features.
Enhances time-series modeling by dropping patches, improving efficiency and adaptability.
A long user history inevitably reflects the transitions of personal interests over time. The analyses on the user history require the robust sequential model to anticipate the transitions and the decays of user interests. The user history is often modeled by various RNN structures, but the RNN structures in the recomme…
New framework improves multivariate time series forecasting by minimizing redundant information.
WS-II algorithm segments trajectories with high accuracy.
ST-GCN improves rs-fMRI prediction accuracy by modeling spatio-temporal graph connectivity.
Deep RL optimizes compiler passes for better performance.