arXiv research

Improved atomistic model predicts molecular properties using weighted skip-connections.

The paper proposes new methods to accurately attribute online advertising revenue.

Novel spectral embedding considers node weights for graph analysis.

Prior-weighted logistic regression has become a standard tool for calibration in speaker recognition. Logistic regression is the optimization of the expected value of the logarithmic scoring rule. We generalize this via a parametric family of proper scoring rules. Our theoretical analysis shows how different members of…

Stability of weighted extremal manifolds proven through blowups.

Weight normalization speeds up matrix sensing problems.

We investigate the $m$-relative entropy, which stems from the Bregman divergence, on weighted Riemannian and Finsler manifolds. We prove that the displacement $K$-convexity of the $m$-relative entropy is equivalent to the combination of the nonnegativity of the weighted Ricci curvature and the $K$-convexity of the weig…

New method finds profitable investment opportunities by considering additional financial variables.

Sketches linear classifiers using Weight-Median Sketch for efficient data stream analysis.

A new model clusters network nodes based on relative edge weights.

This paper gives an exposition of relative weight filtrations on completions of mapping class groups associated to a stable degeneration of marked genus g curves. These relative weight filtrations have been constructed using Galois theory (with Matsumoto) and Hodge theory (with Pearlstein and Terasoma). It is shown tha…

We introduce a class of generalized relative entropies (inspired by the Bregman divergence in information theory) on the Wasserstein space over a weighted Riemannian or Finsler manifold. We prove that the convexity of all the entropies in this class is equivalent to the combination of the nonnegative weighted Ricci cur…

We develop a relative version of Kostant's harmonic theory and use this to prove a relative version of Kostant's theorem on Lie algebra (co)homology. These are associated to two nested parabolic subalgebras in a semisimple Lie algebra. We show how relative homology groups can be used to realize representations with low…

Deep Convolutional Neural Networks (CNNs) are more powerful than Deep Neural Networks (DNN), as they are able to better reduce spectral variation in the input signal. This has also been confirmed experimentally, with CNNs showing improvements in word error rate (WER) between 4-12% relative compared to DNNs across a var…

Researchers classify invariant operators on weighted densities.

Under the assumption of asymptotic relative Chow-stability for polarized algebraic manifolds $(M, L)$, a series of weighted balanced metrics $ω_m$, $m \gg 1$, called polybalanced metrics, are obtained from complete linear systems $|L^m|$ on $M$. Then the asymptotic behavior of the weights as $m \to \infty$ will be stud…

Automates feature selection and weighting in molecular systems.

Paper develops Lie theory for Rota-Baxter operators on Lie algebras and groups.

Paper proposes a fast stochastic algorithm for neural network quantization with error bounds.

Proposes a multi-objective Q-network for dynamic weights in deep RL.

New MCMC method improves sampling from multimodal distributions.

New method constructs relative invariants for group actions on extended manifolds.

New algorithm reduces ERM problem size while maintaining accuracy.

In this paper, we investigate the relative Gel'fand-Kalinin-Fuks cohomology groups of the formal Hamiltonian vector fields on R^4. In the case of formal Hamiltonian vector fields on R^2, we computed the relative Gel'fand-Kalinin-Fuks cohomology groups of weight <20 in the paper by Mikami-Nakae-Kodama. The main strategy…

Analysis shows preference for Chinese yuan in global trade network.

Researchers address the generation of differential invariants for geometric structures.

New model predicts stock performance in large equity markets.

In this paper we study the relative Chow and $K$-stability of toric manifolds in the toric sense. First, we give a criterion for relative $K$-stability and instability of toric Fano manifolds in the toric sense. The reduction of relative Chow stability on toric manifolds will be investigated using the Hibert-Mumford cr…

New algorithm for weighted low rank approximation with provable guarantees.

New tests detect asphericity in complex pairs, simplifying previous proofs.

Proposes a new model for clustering multiplex networks with compositional data.

A new simple method for representing Artin braid groups.

In this paper we prove general inequalities involving the weighted mean curvature of compact submanifolds immersed in weighted manifolds. As a consequence we obtain a relative linear isoperimetric inequality for such submanifolds. We also prove an extrinsic upper bound to the first non zero eigenvalue of the drift Lapl…

New surface area measures defined for ball-convex bodies, leading to entropy and inequalities.

Sparse regularization reduces NLP model complexity without sacrificing accuracy.

Study on convergence of SDEs using entropy methods.

In "The Gel'fand-Kalinin-Fuks class and characteristic classes of transversely symplectic foliations", arXiv:0910.3414, (October 2009) by D.Kotschick and S.Morita, the relative Gel'fand-Kalinin-Fuks cohomology groups of the formal Hamiltonian vector fields without constant vector fields on 2n-plane were characterized b…

Orthogonal initialization speeds up convergence in deep linear networks.

Equivalence found between two types of embeddings for Kähler manifolds.

Algorithm learns Sherrington-Kirkpatrick model parameters at low temperatures.

Post-training quantization saves resources for neural networks.

Attention to entropic communication improves message decoding and cooperation.

A method combines deep learning and G-estimation for causal mediation analysis.

Sparse learning speeds up neural network training without sacrificing accuracy.

We propose a sample-efficient alternative for importance weighting for situations where one only has sample access to the probability distribution that generates the observations. Our new method, called Geometric Resampling (GR), is described and analyzed in the context of online combinatorial optimization under semi-b…

The study analyzes numerical stability in large language models using mixed-precision arithmetic.

A new method uses nearest neighbors for importance weighting.

Equivalence found between certain Kahler and Sasaki metrics.