arXiv research

Constructs a universal average for Lie algebra elements.

Enhances UPSA to reduce noise in financial data.

New analysis shows halting time is predictable for large models, improving optimization efficiency.

Consider a family of portfolio strategies with the aim of achieving the asymptotic growth rate of the best one. The idea behind Cover's universal portfolio is to build a wealth-weighted average which can be viewed as a buy-and-hold portfolio of portfolios. When an optimal portfolio exists, the wealth-weighted average c…

We compute the rational cohomology of the universal family of smooth cubic surfaces using Vassiliev's method of simplicial resolution. Modulo embedding, the universal family has cohomology isomorphic to that of $\mathbb{P}^2$. A consequence of our theorem is that over the finite field $\mathbb{F}_q$, away from finitely…

Under-parameterized networks can either copy or average teacher weights, leading to universal optimal solutions.

We analyze the constituents stocks of the Dow Jones Industrial Average (DJIA30) and the Standard & Poor's 100 index (S&P100) of the NYSE stock exchange market. Surprisingly, we discover the data collapse of the histograms of the DJIA30 price fluctuations and of the S&P100 price fluctuations to the universal non-paramet…

New method evaluates LLMs fairness in universal prediction.

Frame Averaging makes neural networks invariant or equivariant to new symmetries.

A new approach to continuous-time universal portfolios using pathwise Itô calculus.

The common assumption of universal behavior in stock market data can sometimes lead to false conclusions. In statistical physics, the Hurst exponents characterizing long-range correlations are often closely related to universal exponents. We show, that in the case of time series of the traded value, these Hurst exponen…

Novel approach to universal online learning for bounded losses, closing open problems.

DAmageNet generates universal adversarial samples with high transferability.

This paper develops several average-case reduction techniques to show new hardness results for three central high-dimensional statistics problems, implying a statistical-computational gap induced by robustness, a detection-recovery gap and a universality principle for these gaps. A main feature of our approach is to ma…

Solves open problem on universally consistent online learning with unbounded losses.

The paper derives a new theorem for predicting batches of data.

DoWG optimizer automatically adapts to convex and nonsmooth problems without tuning.

This note provides a neat and enjoyable expansion and application of the magnificent Ordentlich-Cover theory of "universal portfolios." I generalize Cover's benchmark of the best constant-rebalanced portfolio (or 1-linear trading strategy) in hindsight by considering the best bilinear trading strategy determined in hin…

Improved algorithm speeds up generation of universal adversarial perturbations.

UTOPIA aggregates multiple prediction intervals efficiently.

We study the average shape of a fluctuation of a time series x(t), that is the average value <x(t)-x(0)>_T before x(t) first returns, at time T, to its initial value x(0). For large classes of stochastic processes we find that a scaling law of the form <x(t) - x(0)>_T = T^αf(t/T) is obeyed. The scaling function f(s) is…

Generalizes causal inference to high-dimensional outcomes.

Reduces bounded loss learning to binary classification.

Let $(M, g)$ be a complete non-compact Kähler manifold with non-negative and bounded holomorphic bisectional curvature. Extending our techniques developed in \cite{CT3}, we prove that the universal cover $\wt M$ of $M$ is biholomorphic to $\ce^n$ provided either that $(M, g)$ has average quadratic curvature decay, or $…

Although stochastic gradient descent (SGD) method and its variants (e.g., stochastic momentum methods, AdaGrad) are the choice of algorithms for solving non-convex problems (especially deep learning), there still remain big gaps between the theory and the practice with many questions unresolved. For example, there is s…

Given two or more Deep Neural Networks (DNNs) with the same or similar architectures, and trained on the same dataset, but trained with different solvers, parameters, hyper-parameters, regularization, etc., can we predict which DNN will have the best test accuracy, and can we do so without peeking at the test data? In …

Study evaluates the impact of academic support center's face-to-face assistance on student performance.

We study power-law correlations properties of the Google search queries for Dow Jones Industrial Average (DJIA) component stocks. Examining the daily data of the searched terms with a combination of the rescaled range and rescaled variance tests together with the detrended fluctuation analysis, we show that the searche…

Signatures of universality are detected by comparing individual eigenvalue distributions and level spacings from financial covariance matrices to random matrix predictions. A chopping procedure is devised in order to produce a statistical ensemble of asset-price covariances from a single instance of financial data sets…

The universal perturbative invariants of rational homology spheres can be extracted from the Chern-Simons partition function by combining perturbative and nonperturbative results. We spell out the general procedure to compute these invariants, and we work out in detail the case of Seifert spaces. By extending some prev…

This work initiates a general study of learning and generalization without the i.i.d. assumption, starting from first principles. While the traditional approach to statistical learning theory typically relies on standard assumptions from probability theory (e.g., i.i.d. or stationary ergodic), in this work we are inter…

A new method forecasts financial tail risks by combining and weighting quantiles.

The universe's shape and size are determined in general cosmological models.

This paper explores the limits of deep learning in poly-time.

Study confirms the square-root law in price impact across Tokyo stocks.

We conclude from an analysis of high resolution NYSE data that the distribution of the traded value $f_i$ (or volume) has a finite variance $σ_i$ for the very large majority of stocks $i$, and the distribution itself is non-universal across stocks. The Hurst exponent of the same time series displays a crossover from we…

New averaging technique speeds up Newton method convergence.

Deep Convolutional Networks (DCNs) have been shown to be vulnerable to adversarial examples---perturbed inputs specifically designed to produce intentional errors in the learning algorithms at test time. Existing input-agnostic adversarial perturbations exhibit interesting visual patterns that are currently unexplained…

No universal trading strategy exists due to mathematical impossibilities.

Using a large-scale Deep Learning approach applied to a high-frequency database containing billions of electronic market quotes and transactions for US equities, we uncover nonparametric evidence for the existence of a universal and stationary price formation mechanism relating the dynamics of supply and demand for a s…

Diversified risk parity strategies outperform equally-weighted portfolios in various asset universes.

Max-margin classifiers' behavior is studied in high dimensions with non-Gaussian features.

How many bits of information are revealed by a learning algorithm for a concept class of VC-dimension $d$? Previous works have shown that even for $d=1$ the amount of information may be unbounded (tend to $\infty$ with the universe size). Can it be that all concepts in the class require leaking a large amount of inform…

Study on stochastic approximation with Polyak-Ruppert averaging for linear systems.

We analyze a new spectral graph matching algorithm, GRAph Matching by Pairwise eigen-Alignments (GRAMPA), for recovering the latent vertex correspondence between two unlabeled, edge-correlated weighted graphs. Extending the exact recovery guarantees established in the companion paper for Gaussian weights, in this work,…

New insights into bias mitigation show DRO isn't a complete solution.

The study examines averages of Laplacian determinants over large genus moduli spaces.

We study geodesics on the modular surface, comparing WP and hyperbolic metrics.