arXiv research

SLIP secures LLMs on edge devices by splitting computation and protecting sensitive parts.

DeepPeep attacks DNN architectures to reveal design details, posing IP theft risks.

We consider a fundamental integer programming (IP) model for cost-benefit analysis flood protection through dike building in the Netherlands, due to Verweij and Zwaneveld. Experimental analysis with data for the Ijsselmeer lead to integral optimal solution of the linear programming relaxation of the IP model. This natu…

Training machine learning (ML) models is expensive in terms of computational power, amounts of labeled data and human expertise. Thus, ML models constitute intellectual property (IP) and business value for their owners. Embedding digital watermarks during model training allows a model owner to later identify their mode…

The study improves off-policy learning by smoothing IPS and provides a generalization bound.

In many developing countries intellectual property infringement and the commerce of pirate goods is an entrepreneurial activity. Digital piracy is very often the only media for having access to music, cinema, books and software. At the same time, bio-prospecting and infringement of indigenous knowledge rights by intern…

New method tunes prior IP to data for flexible predictive distributions.

Integer programming (IP) is a general optimization framework widely applicable to a variety of unstructured and structured problems arising in, e.g., scheduling, production planning, and graph optimization. As IP models many provably hard to solve problems, modern IP solvers rely on many heuristics. These heuristics ar…

DVIP improves on IP-based methods by using IPs as priors over latent functions.

The commercialization of deep learning creates a compelling need for intellectual property (IP) protection. Deep neural network (DNN) watermarking has been proposed as a promising tool to help model owners prove ownership and fight piracy. A popular approach of watermarking is to train a DNN to recognize images with ce…

This paper revisits the classic iterative proportional scaling (IPS) from a modern optimization perspective. In contrast to the criticisms made in the literature, we show that based on a coordinate descent characterization, IPS can be slightly modified to deliver coefficient estimates, and from a majorization-minimizat…

We consider the representation power of siamese-style similarity functions used in neural network-based graph embedding. The inner product similarity (IPS) with feature vectors computed via neural networks is commonly used for representing the strength of association between two nodes. However, only a little work has b…

Study generalizes property elicitation to imprecise probabilities.

Extends V-IP framework to use LLMs for generating task-relevant concepts, improving interpretability and performance.

MEC-IP uses IP to efficiently find MECs in BNs from observational data.

A new method for making interpretable predictions by sequentially asking questions, faster and more efficient.

A Riemannian manifold is called IP, if the eigenvalues of its skew-symmetric curvature operator are pointwise constant. It was previously shown that for all n\ge 4, except n=7, any IP manifold either has constant curvature, or is a warped product, with some specific function, of a line and a space of constant curvature…

C-IP improves LLMs' query selection for interactive tasks by estimating uncertainty robustly.

Let M be a pseudo-Riemannian manifold with a pseudo-Hermitian complex structure $J$. We give necessary and sufficient conditions that the curvature operator $R(π)$ is complex linear when $π$ is a $J$ invariant real 2 plane. Under this assumption, we study when M is complex IP - i.e. the spectrum, or more generally the …

Paper designs optimal ECOCs using IP for robust multiclass classification.

Paper proposes CounterSample to improve convergence in LTR models.

On the Geroch-Kronheimer-Penrose future completion $IP(X)$ of a spacetime $X$, there are two frequently used topologies. We systematically examine $τ_+$, the stronger (metrizable) of them, which is the coarsest causally continuous topology, obtaining a variety of novel results, among them a complete characterization of…

New sampling method uses gradient-free IPS with RKHS velocity field.

We introduce the implicit processes (IPs), a stochastic process that places implicitly defined multivariate distributions over any finite collections of random variables. IPs are therefore highly flexible implicit priors over functions, with examples including data simulators, Bayesian neural networks and non-linear tr…

The Information Plane theory predicts autoencoders do not compress input information.

We construct a family of pseudo-Riemannian manifolds so that the skew-symmetric curvature operator, the Jacobi operator, and the Szabo operator have constant eigenvalues on their domains of definition. This provides new and non-trivial examples of Osserman, Szabo, and IP manifolds. We also study when the associated Jor…

Analyzing deep neural networks (DNNs) via information plane (IP) theory has gained tremendous attention recently as a tool to gain insight into, among others, their generalization ability. However, it is by no means obvious how to estimate mutual information (MI) between each hidden layer and the input/desired output, …

We describe two applications of machine learning in the context of IP/Optical networks. The first one allows agile management of resources at a core IP/Optical network by using machine learning for short-term and long-term prediction of traffic flows and joint global optimization of IP and optical layers using colorles…

Method identifies IPS governing equations from particle data efficiently.

In this small article one compromise monetization strategy is proposed, which hopefully may lead to a more satisfactory coexistence of IP manufacturers and consumers. The motto is "fair exchange": you use our IP-product, we use your product (in form of money); when you do not need our product any more, we change back.

Conformal Prediction Regions match Imprecise Highest Density Regions under consonance.

Pessimistic estimator improves multi-objective policy optimization.

AIPS improves ranking policy evaluation by adapting to diverse user behavior.

Improved CAEs reduce training time and enhance generalization.

Adaptive selection of IPs improves online GP performance.

This paper investigates the statistical properties of within-country GDP and industrial production (IP) growth rate distributions. Many empirical contributions have recently pointed out that cross-section growth rates of firms, industries and countries all follow Laplace distributions. In this work, we test whether als…

Paper tackles BNSL with IP, improving quality of solutions.

We propose shifted inner-product similarity (SIPS), which is a novel yet very simple extension of the ordinary inner-product similarity (IPS) for neural-network based graph embedding (GE). In contrast to IPS, that is limited to approximating positive-definite (PD) similarities, SIPS goes beyond the limitation by introd…

This work introduces a new metric for comparing imprecise probability models.

Comparing with traditional learning criteria, such as mean square error (MSE), the minimum error entropy (MEE) criterion is superior in nonlinear and non-Gaussian signal processing and machine learning. The argument of the logarithm in Renyis entropy estimator, called information potential (IP), is a popular MEE cost i…

Contextual bandit methods fail with deficient support data.

Topological Quantum Field Theories (TQFTs) pertinent to some emergent low energy phenomena of condensed matter lattice models in 2+1 and 3+1D are explored. Many of our field theories are highly-interacting without free quadratic analogs. Some of our bosonic TQFTs can be regarded as the continuum field theory formulatio…

A new estimator reduces variance in slate bandit OPE.

Adaptive IP approach optimizes intervention design for causal graph recovery.

We study the off-policy evaluation problem---estimating the value of a target policy using data collected by another policy---under the contextual bandit model. We consider the general (agnostic) setting without access to a consistent model of rewards and establish a minimax lower bound on the mean squared error (MSE).…

A pseudo-Riemannian manifold is said to be spacelike Jordan IP if the Jordan normal form of the skew-symmetric curvature operator depends upon the point of the manifold, but not upon the particular spacelike 2-plane in the tangent bundle at that point. We use methods of algebraic topology to classify connected spacelik…

Unified framework solves nonlinear PDEs and IPs using Gaussian processes.

We study offline recommender learning from explicit rating feedback in the presence of selection bias. A current promising solution for the bias is the inverse propensity score (IPS) estimation. However, the performance of existing propensity-based methods can suffer significantly from the propensity estimation bias. I…