arXiv research

New algorithm for duelling bandits with weak regret in adversarial settings.

Study evaluates new models using human feedback from another model.

Simple uniqueness proof for McKean-Vlasov systems with weak feedback.

Online boosting for multiclass classification with limited feedback.

Interactive weak supervision learns useful heuristics from user feedback.

New framework transfers latent knowledge from weak to strong models.

Online boosting for multilabel ranking with limited feedback.

This paper studies the Glosten Milgrom model whose risky asset value admits an arbitrary discrete distribution. Contrast to existing results on insider's models, the insider's optimal strategy in this model, if exists, is not of feedback type. Therefore a weak formulation of equilibrium is proposed. In this weak formul…

Framework identifies population quantities from MNAR feedback using weak shadow variables from pretrained models.

Efficient algorithms for online multiclass linear classification with bandit feedback under linear separability conditions.

Novel ramp loss method improves weakly supervised machine translation and parsing.

Formalizes weak and strong verification for LLMs, controlling errors without assumptions.

Boosting for off-policy learning reduces empirical risk.

New graph feedback model for bandits with improved regret bounds.

The paper constrains families of smooth 4-manifolds using Seiberg-Witten invariants.

In this manuscript we analyse the leading statistical properties of fluctuations of (log) 3-month US Treasury bill quotation in the secondary market, namely: probability density function, autocorrelation, absolute values autocorrelation, and absolute values persistency. We verify that this financial instrument, in spit…

This paper extends a Kyle model to include price-responsive traders, revealing new dynamics and equilibria.

Model shows liquidity crises linked to past volatility and price trends.

Proposes a model to classify nodes in networks using weighted feedback relations.

We classify isotopy classes of irreducible Heegaard splittings of solvmanifolds. If the monodromy of the solvmanifold can be expressed as a 2 x 2 matrix with 0 in the lower right hand corner (as always is true when the absolute value of the trace is 3), then any irreducible splitting is strongly irreducible and of genu…

Proposes estimators for complex dose-response curves using kernel methods.

Neural networks learn distance-based representations, not just intensity.

We study principal curvatures of fibers and Heegaard surfaces smoothly embedded in hyperbolic 3-manifolds. It is well known that a fiber or a Heegaard surface in a hyperbolic 3-manifold cannot have principal curvatures everywhere less than one in absolute value. We show that given an upper bound on the genus of a minim…

We explore whether useful temporal neural generative models can be learned from sequential data without back-propagation through time. We investigate the viability of a more neurocognitively-grounded approach in the context of unsupervised generative modeling of sequences. Specifically, we build on the concept of predi…

Algorithm learns to bid optimally in repeated first-price auctions with censored feedback.

PDO optimizes LLM prompts without labels, improving performance.

Learning the right graph representation from noisy, multisource data has garnered significant interest in recent years. A central tenet of this problem is relational learning. Here the objective is to incorporate the partial information each data source gives us in a way that captures the true underlying relationships.…

We decompose, within an ARCH framework, the daily volatility of stocks into overnight and intra-day contributions. We find, as perhaps expected, that the overnight and intra-day returns behave completely differently. For example, while past intra-day returns affect equally the future intra-day and overnight volatilitie…

Model user preferences for conversational LLMs using weak rewards.

There is increasing interest in learning algorithms that involve interaction between human and machine. Comparison-based queries are among the most natural ways to get feedback from humans. A challenge in designing comparison-based interactive learning algorithms is coping with noisy answers. The most common fix is to …

Algorithm minimizes regret in dueling bandits with contextualized utilities.

Algorithm minimizes regret in non-stationary dueling bandits with unknown parameters.

Improved mistake bound for group linear separable cases in online multiclass linear classification.

Boosting improves online decision-making for large expert sets.

A challenge in training discriminative models like neural networks is obtaining enough labeled training data. Recent approaches use generative models to combine weak supervision sources, like user-defined heuristics or knowledge bases, to label training data. Prior work has explored learning accuracies for these source…

xAI-GAN improves GANs by providing richer feedback, enhancing image quality.

Thompson Sampling tackles USS, a sequential selection problem without feedback.

We present an efficient algorithm for calculating the number of components of an integral lamination on an $n$-punctured disk, given its Dynnikov coordinates. The algorithm requires $O(n^2M)$ arithmetic operations, where $M$ is the sum of the absolute values of the Dynnikov coordinates.

We prove that for n>4 there is no compact arithmetic hyperbolic n-manifold whose Euler characteristic has absolute value equal to 2. In particular, this shows the nonexistence of arithmetically defined hyperbolic rational homology n-sphere with n even different than 4.

We propose a method to build quantum memristors in quantum photonic platforms. We firstly design an effective beam splitter, which is tunable in real-time, by means of a Mach-Zehnder-type array with two equal 50:50 beam splitters and a tunable retarder, which allows us to control its reflectivity. Then, we show that th…

Recurrent neural networks (RNNs) are notoriously difficult to train. When the eigenvalues of the hidden to hidden weight matrix deviate from absolute value 1, optimization becomes difficult due to the well studied issue of vanishing and exploding gradients, especially when trying to learn long-term dependencies. To cir…

Establishes a link between risk measures and uniform integrability in finance.

Active learning improves inspection systems by using weakly labeled data.

MallowsPO enhances LLM fine-tuning with a dispersion index of human preferences.

The study proves knots and certain links support taut foliations.

In this paper, we study the power of Gaussian curvature flow of a compact convex hypersurface and establish its Harnack inequality when the power is negative. In the Harnack inequality, we require that the absolute value of the power is strictly positive and strictly less than the inverse of the dimension of the hypers…

A finitely generated module over the ring L=Z[t, t^{-1}] of integer Laurent polynomials that has no Z-torsion is determined by a pair of sub-lattices of L^d. Their indices are the absolute values of the leading and trailing coefficients of the order of the module. This description has applications in knot theory.

A complex-valued convolutional network (convnet) implements the repeated application of the following composition of three operations, recursively applying the composition to an input vector of nonnegative real numbers: (1) convolution with complex-valued vectors followed by (2) taking the absolute value of every entry…