Source author record

Ragnar Thobaben

Ragnar Thobaben appears in the imported research catalog. Authorship, coauthor and topic links are available while profile ownership is still unclaimed.

ResearcherUnclaimed source record

Information Theory math.IT Machine Learning math.OC Systems and Control

Catalog footprint

What is connected

12works

5topics

4close collaborators

Actions

Connect this record

Open graph Browse works

Inspect adjacent papers, topics, institutions and collaborators without losing the researcher page.

Building this map preview

BZPEER is loading the nearby papers, people, topics and institutions for this page.

preprint2026arXiv

A Hierarchical Sampling Framework for bounding the Generalization Error of Federated Learning

We study expected generalization bounds for the Hierarchical Federated Learning (HFL) setup using Wasserstein distance. We introduce a generalized framework in which data is sampled hierarchically, and we model it with a multi-layered tree structure that induces dependencies among the clients' datasets. We derive generalization bounds in terms of Wasserstein distance under the Lipschitz assumption on the loss function, by applying a supersample construction that allows us to measure the sensitivity of the algorithm to the change of a single node in the sampling tree. By leveraging the FL structure, we recover and strictly imply existing state-of-the-art conditional mutual information (CMI) bounds in the case of bounded losses. We also show that our bound can be applied together with Differential Privacy assumptions, to recover generalization bounds based on algorithmic privacy. To assess the tightness of our bounds, we study the Gaussian Location Model (GLM) and show that we recover the actual asymptotic rate of the generalization error.

preprint2026arXiv

Density-Ratio Losses for Post-Hoc Learning to Defer

We study post-hoc Learning to Defer (L2D) through the lens of ideal distributions: divergence-regularized reweightings of the data distribution under which a model attains low loss. We define deferral via the density-ratio between a model's and an expert's ideals. Using the reduction from density-ratio estimation to class-probability estimation, we derive the DR CPE losses for post-hoc L2D scorers. Deferral decisions are then made by thresholding the scorer, allowing deferral rates to be adjusted without retraining. For KL-based ideal distributions, our deferral rules recovers Chow's rule under the original distribution and a connection to an expert-tilted Bayes posterior -- which incorporates the expert's performance -- depending on if the ideal distributions are joint or marginal distributions. Experimentally, our approach is competitive compared to common baselines and more robust across dataset settings. More broadly, our results cast post-hoc L2D as density-ratio learning between ideal distributions, bridging Chow-style rules, expert comparison, and elucidating connections to related learning settings including anomaly detection.

preprint2022arXiv

Tighter expected generalization error bounds via Wasserstein distance

This work presents several expected generalization error bounds based on the Wasserstein distance. More specifically, it introduces full-dataset, single-letter, and random-subset bounds, and their analogues in the randomized subsample setting from Steinke and Zakynthinou [1]. Moreover, when the loss function is bounded and the geometry of the space is ignored by the choice of the metric in the Wasserstein distance, these bounds recover from below (and thus, are tighter than) current bounds based on the relative entropy. In particular, they generate new, non-vacuous bounds based on the relative entropy. Therefore, these results can be seen as a bridge between works that account for the geometry of the hypothesis space and those based on the relative entropy, which is agnostic to such geometry. Furthermore, it is shown how to produce various new bounds based on different information measures (e.g., the lautum information or several $f$-divergences) based on these bounds and how to derive similar bounds with respect to the backward channel using the presented proof techniques.

preprint2021arXiv

Quadratic Signaling Games with Channel Combining Ratio

In this study, Nash and Stackelberg equilibria of single-stage and multi-stage quadratic signaling games between an encoder and a decoder are investigated. In the considered setup, the objective functions of the encoder and the decoder are misaligned, there is a noisy channel between the encoder and the decoder, the encoder has a soft power constraint, and the decoder has also noisy observation of the source to be estimated. We show that there exist only linear encoding and decoding strategies at the Stackelberg equilibrium, and derive the equilibrium strategies and costs. Regarding the Nash equilibrium, we explicitly characterize affine equilibria for the single-stage setup and show that the optimal encoder (resp. decoder) is affine for an affine decoder (resp. encoder) for the multi-stage setup. For the decoder side, between the information coming from the encoder and noisy observation of the source, our results describe what should be the combining ratio of these two channels. Regarding the encoder, we derive the conditions under which it is meaningful to transmit a message.

preprint2020arXiv

Region-based Energy Neural Network for Approximate Inference

Region-based free energy was originally proposed for generalized belief propagation (GBP) to improve loopy belief propagation (loopy BP). In this paper, we propose a neural network based energy model for inference in general Markov random fields (MRFs), which directly minimizes the region-based free energy defined on region graphs. We term our model Region-based Energy Neural Network (RENN). Unlike message-passing algorithms, RENN avoids iterative message propagation and is faster. Also different from recent deep neural network based models, inference by RENN does not require sampling, and RENN works on general MRFs. RENN can also be employed for MRF learning. Our experiments on marginal distribution estimation, partition function estimation, and learning of MRFs show that RENN outperforms the mean field method, loopy BP, GBP, and the state-of-the-art neural network based model.

preprint2020arXiv

The Convex Information Bottleneck Lagrangian

The information bottleneck (IB) problem tackles the issue of obtaining relevant compressed representations $T$ of some random variable $X$ for the task of predicting $Y$. It is defined as a constrained optimization problem which maximizes the information the representation has about the task, $I(T;Y)$, while ensuring that a certain level of compression $r$ is achieved (i.e., $ I(X;T) \leq r$). For practical reasons, the problem is usually solved by maximizing the IB Lagrangian (i.e., $\mathcal{L}_{\text{IB}}(T;β) = I(T;Y) - βI(X;T)$) for many values of $β\in [0,1]$. Then, the curve of maximal $I(T;Y)$ for a given $I(X;T)$ is drawn and a representation with the desired predictability and compression is selected. It is known when $Y$ is a deterministic function of $X$, the IB curve cannot be explored and another Lagrangian has been proposed to tackle this problem: the squared IB Lagrangian: $\mathcal{L}_{\text{sq-IB}}(T;β_{\text{sq}})=I(T;Y)-β_{\text{sq}}I(X;T)^2$. In this paper, we (i) present a general family of Lagrangians which allow for the exploration of the IB curve in all scenarios; (ii) provide the exact one-to-one mapping between the Lagrange multiplier and the desired compression rate $r$ for known IB curve shapes; and (iii) show we can approximately obtain a specific compression level with the convex IB Lagrangian for both known and unknown IB curve shapes. This eliminates the burden of solving the optimization problem for many values of the Lagrange multiplier. That is, we prove that we can solve the original constrained problem with a single optimization.

preprint2016arXiv

Multi-Phase Smart Relaying and Cooperative Jamming in Secure Cognitive Radio Networks

In this paper we investigate cooperative secure communications in a four-node cognitive radio network where the secondary receiver is treated as a potential eavesdropper with respect to the primary transmission. The secondary user is allowed to transmit his own signals under the condition that the primary user's secrecy rate and transmission scheme are intact. Under this setting we derive the secondary user's achievable rates and the related constraints to guarantee the primary user's weak secrecy rate, when Gelfand-Pinsker coding is used at the secondary transmitter. In addition, we propose a multi-phase transmission scheme to include 1) the phases of the clean relaying with cooperative jamming and 2) the latency to successfully decode the primary message at the secondary transmitter. A capacity upper bound for the secondary user is also derived. Numerical results show that: 1) the proposed scheme can outperform the traditional ones by properly selecting the secondary user's parameters of different transmission schemes according to the relative positions of the nodes; 2) the derived capacity upper bound is close to the secondary user's achievable rate within 0.3 bits/channel use, especially when the secondary transmitter/receiver is far/close enough to the primary receiver/transmitter, respectively. Thereby, a smart secondary transmitter is able to adapt its relaying and cooperative jamming to guarantee primary secrecy rates and to transmit its own data at the same time from relevant geometric positions.

preprint2015arXiv

Differentially Private State Estimation in Distribution Networks with Smart Meters

State estimation is routinely being performed in high-voltage power transmission grids in order to assist in operation and to detect faulty equipment. In low- and medium-voltage power distribution grids, on the other hand, few real-time measurements are traditionally available, and operation is often conducted based on predicted and historical data. Today, in many parts of the world, smart meters have been deployed at many customers, and their measurements could in principle be shared with the operators in real time to enable improved state estimation. However, customers may feel reluctance in doing so due to privacy concerns. We therefore propose state estimation schemes for a distribution grid model, which ensure differential privacy to the customers. In particular, the state estimation schemes optimize different performance criteria, and a trade-off between a lower bound on the estimation performance versus the customers' differential privacy is derived. The proposed framework is general enough to be applicable also to other distribution networks, such as water and gas networks.

preprint2014arXiv

Communication and Interference Coordination

We study the problem of controlling the interference created to an external observer by a communication processes. We model the interference in terms of its type (empirical distribution), and we analyze the consequences of placing constraints on the admissible type. Considering a single interfering link, we characterize the communication-interference capacity region. Then, we look at a scenario where the interference is jointly created by two users allowed to coordinate their actions prior to transmission. In this case, the trade-off involves communication and interference as well as coordination. We establish an achievable communication-interference region and show that efficiency is significantly improved by coordination.

preprint2011arXiv

Bilayer LDPC Convolutional Codes for Half-Duplex Relay Channels

In this paper we present regular bilayer LDPC convolutional codes for half-duplex relay channels. For the binary erasure relay channel, we prove that the proposed code construction achieves the capacities for the source-relay link and the source-destination link provided that the channel conditions are known when designing the code. Meanwhile, this code enables the highest transmission rate with decode-and-forward relaying. In addition, its regular degree distributions can easily be computed from the channel parameters, which significantly simplifies the code optimization. Numerical results are provided for both binary erasure channels (BEC) and AWGN channels. In BECs, we can observe that the gaps between the decoding thresholds and the Shannon limits are impressively small. In AWGN channels, the bilayer LDPC convolutional code clearly outperforms its block code counterpart in terms of bit error rate.

preprint2011arXiv

Performance Analysis and Design of Two Edge Type LDPC Codes for the BEC Wiretap Channel

We consider transmission over a wiretap channel where both the main channel and the wiretapper's channel are Binary Erasure Channels (BEC). We propose a code construction method using two edge type LDPC codes based on the coset encoding scheme. Using a standard LDPC ensemble with a given threshold over the BEC, we give a construction for a two edge type LDPC ensemble with the same threshold. If the given standard LDPC ensemble has degree two variable nodes, our construction gives rise to degree one variable nodes in the code used over the main channel. This results in zero threshold over the main channel. In order to circumvent this problem, we numerically optimize the degree distribution of the two edge type LDPC ensemble. We find that the resulting ensembles are able to perform close to the boundary of the rate-equivocation region of the wiretap channel. There are two performance criteria for a coding scheme used over a wiretap channel: reliability and secrecy. The reliability measure corresponds to the probability of decoding error for the intended receiver. This can be easily measured using density evolution recursion. However, it is more challenging to characterize secrecy, corresponding to the equivocation of the message for the wiretapper. Méasson, Montanari, and Urbanke have shown how the equivocation can be measured for a broad range of standard LDPC ensembles for transmission over the BEC under the point-to-point setup. By generalizing the method of Méasson, Montanari, and Urbanke to two edge type LDPC ensembles, we show how the equivocation for the wiretapper can be computed. We find that relatively simple constructions give very good secrecy performance and are close to the secrecy capacity. However finding explicit sequences of two edge type LDPC ensembles which achieve secrecy capacity is a more difficult problem. We pose it as an interesting open problem.

preprint2010arXiv

Nested Polar Codes for Wiretap and Relay Channels

We show that polar codes asymptotically achieve the whole capacity-equivocation region for the wiretap channel when the wiretapper's channel is degraded with respect to the main channel, and the weak secrecy notion is used. Our coding scheme also achieves the capacity of the physically degraded receiver-orthogonal relay channel. We show simulation results for moderate block length for the binary erasure wiretap channel, comparing polar codes and two edge type LDPC codes.

Institution

Affiliation not imported yet

This author record came from a source that does not expose affiliation metadata. Once the author claims the profile or we enrich the record from another provider, this section will link to the concrete institution.

Topic footprint