arXiv research

Deep neural networks optimize inventory decisions in complex supply chains.

We study in this paper the remnants of the contact partial order on the orbits of the adjoint action of contactomorphism groups on their Lie algebras. Our main interest is a class of non-compact contact manifolds, called convex at infinity.

Consider cotangent bundles of exotic spheres, with their canonical symplectic structure. They admit automorphisms which preserve the part at infinity of one fibre, and which are analogous to the square of a Dehn twist. Pursuing that analogy, we show that they have infinite order up to isotopy (inside the group of all a…

We develop a systematic method for classifying supersymmetric orbifold compactifications of M-theory. By restricting our attention to abelian orbifolds with low order, in the special cases where elements do not include coordinate shifts, we construct a "periodic table" of such compactifications, organized according to …

According to the work of Laitinen, Morimoto, Oliver and Pawałowski, a finite group $G$ has a smooth effective one fixed point action on some sphere if and only if $G$ is an Oliver group. For some finite Oliver groups $G$ of order up to $216$, and for $G=A_5\times C_n$ for $n=3,5,7$, we present a strategy of excluding o…

The non-gaussianity of processes observed in financial markets and relatively good performance of gaussian models can be reconciled by replacing the Brownian motion with Levy processes whose Levy densities decay as exp(-lambda|x|) or faster, where lambda>0 is large. This leads to asymptotic pricing models. The leading …

Study finds maximal linklessly embeddable graphs up to 11 vertices and their complements.

Kulkarni showed that, if g is greater than 3, a periodic map on an oriented surface S_g of genus g with order more than or equal to 4g is uniquely determined by its order, up to conjugation and power. In this paper, we show that, if g is greater than 30, the same phenomenon happens for periodic maps on the surfaces wit…

Paper improves confidence set construction for SGD using multiplier bootstrap.

The paper characterizes Alexander quandles of finite groups.

We formulate and discuss two conjectures concerning recursive formulae for Branson's $Q$-curvatures. The proposed formulae describe all $Q$-curvatures on manifolds of all even dimensions in terms of respective lower order $Q$-curvatures and lower order GJMS-operators. They are universal in the dimension of the underlyi…

Novel M-theory approach classifies topological phases of matter.

Paper improves CLT and bootstrap approximations for LSA with decreasing step size.

The paper analyzes Q-learning convergence rates with asynchronous updates.

Equity auctions show linear price impact up to a large volume, then non-linear.

New method stabilizes probabilistic ODE solvers for high accuracy.

It is known that the spectrum of the Laplace operator on functions of a closed Riemannian manifold does not determine the integrals of the individual fourth order curvature invariants $\operatorname{scal}^2$, $|\operatorname{ric}|^2$, $|R|^2$, which appear as summands in the second heat invariant $a_2$. We study the an…

Online minimization of an unknown convex function over the interval $[0,1]$ is considered under first-order stochastic bandit feedback, which returns a random realization of the gradient of the function at each query point. Without knowing the distribution of the random gradients, a learning algorithm sequentially choo…

Derives scaling limits and fluctuations for SGD in high dimensions.

Random projections improve classifier generalization without needing to choose the best threshold.

Optimal hidden-target learning for online inventory optimization on general convex sets.

Efficiently estimates mean of symmetric distributions without moments.

Characterizes graphs with leveled embeddings and introduces new graph invariants.

This paper analyzes user-level local differential privacy in distributed systems.

Music FaderNets learns high-level musical qualities from low-level attributes.

Deep reinforcement learning (RL) has shown impressive results in a variety of domains, learning directly from high-dimensional sensory streams. However, when neural networks are trained in a fixed environment, such as a single level in a video game, they will usually overfit and fail to generalize to new levels. When R…

The paper discovers and evaluates support and resistance levels in financial time series.

Minimal graph level sets are concave if boundary is concave.

BILBO optimizes bilevel problems without repeated lower-level optimizations.

We applied Generative Adversarial Networks (GANs) to learn a model of DOOM levels from human-designed content. Initially, we analysed the levels and extracted several topological features. Then, for each level, we extracted a set of images identifying the occupied area, the height map, the walls, and the position of ga…

The paper tackles multi-level fairness in algorithmic systems, addressing bias at both individual and structural levels.

This paper reports the performances of shallow word-level convolutional neural networks (CNN), our earlier work (2015), on the eight datasets with relatively large training data that were used for testing the very deep character-level CNN in Conneau et al. (2016). Our findings are as follows. The shallow word-level CNN…

FRESH combines patient-level and aggregate-level data for better clinical decision making.

New framework tackles bi-level optimization without LLS condition.

This paper improves level generation using VAEs for coherent, logically following segments.

GMVAEs cluster and generate game levels without labels.

TOAD-GAN generates coherent game levels from a single example.

\textit{Multiple Instance Learning} (MIL) is concerned with learning from bags of instances, where only bag labels are given and instance labels are unknown. Existent approaches in this field were mainly designed for the bag-level label prediction (predict labels for bags) but not the instance-level (predict labels for…

Paper explores generalization of AID-based bi-level optimization methods.

The paper calculates dimensions of higher Landau levels on compact manifolds.

Study compares hypergraph and graph-level models for higher-order relational learning.

For leveled spatial graphs, we find a surface embedding that allows cellular embedding.

Researchers infer firm-level supply chain networks from sector-level data to assess systemic risk.

This paper shows how path spaces on two-level manifolds can be Hilbert manifold structures.

New scoring rules for multivariate distributions and level sets.

Bi-GNN models drug interactions using a bi-level graph approach.

We present a novel solution to the problem of simulation-to-real transfer, which builds on recent advances in robot skill decomposition. Rather than focusing on minimizing the simulation-reality gap, we learn a set of diverse policies that are parameterized in a way that makes them easily reusable. This diversity and p…

Same cohomology for curves with levels, proving stable range.