Hyperspherical Prototypical Learning (HPL) is a supervised approach to representation learning that designs class prototypes on the unit hypersphere. The prototypes bias the representations to class separation in a scale invariant and known geometry. Previous approaches to HPL have either of the following shortcomings: (i) they follow an unprincipled optimisation procedure; or (ii) they are theoretically sound, but are constrained to only one possible latent dimension. In this paper, we address both shortcomings. To address (i), we present a principled optimisation procedure whose solution we show is optimal. To address (ii), we construct well-separated prototypes in a wide range of dimensions using linear block codes. Additionally, we give a full characterisation of the optimal prototype placement in terms of achievable and converse bounds, showing that our proposed methods are near-optimal.

Hyperspherical Prototypical Learning (HPL) is a supervised approach to representation learning that designs class prototypes on the unit hypersphere. The prototypes bias the representations to class separation in a scale invariant and known geometry. Previous approaches to HPL have either of the following shortcomings: (i) they follow an unprincipled optimisation procedure; or (ii) they are theoretically sound, but are constrained to only one possible latent dimension. In this paper, we address both shortcomings. To address (i), we present a principled optimisation procedure whose solution we show is optimal. To address (ii), we construct well-separated prototypes in a wide range of dimensions using linear block codes. Additionally, we give a full characterisation of the optimal prototype placement in terms of achievable and converse bounds, showing that our proposed methods are near-optimal.

In this paper we propose an unpredictability-based jamming defense framework based on the principles of Moving Target Defense for a wireless communication problem. Taking advantage of the complex nature of large-scale cyber-physical systems, we consider a platform consisting of a single receiving component but multiple potential transmitting components, each equipped with a multi-antenna phased array. We formulate an optimization problem over the probability simplex that characterizes a randomized receiving angle which seeks to balance between the estimated performance of the transmission and an entropy-based unpredictability measure. Furthermore, we explore the effect of an intelligent adversary that has knowledge of the derived probabilities and optimally places a single-antenna jamming device to disrupt the communication links. Finally, simulation results showcase the efficacy of the proposed algorithm.

This paper studies the truncation method from Alquier [1] to derive high-probability PAC-Bayes bounds for unbounded losses with heavy tails. Assuming that the p-th moment is bounded, the resulting bounds interpolate between a slow rate 1/√n when p=2, and a fast rate 1/n when p→∞ and the loss is essentially bounded. Moreover, the paper derives a high-probability PAC-Bayes bound for losses with a bounded variance. This bound has an exponentially better dependence on the confidence parameter and the dependency measure than previous bounds in the literature. Finally, the paper extends all results to guarantees in expectation and single-draw PAC-Bayes. In order to so, it obtains analogues of the PAC-Bayes fast rate bound for bounded losses from [2] in these settings. The full version of the paper can be found in https://arxiv.org/abs/2403.16681.

In this paper, we present new high-probability PAC-Bayes bounds for different types of losses. Firstly, for losses with a bounded range, we recover a strengthened version of Catoni's bound that holds uniformly for all parameter values. This leads to new fast-rate and mixed-rate bounds that are interpretable and tighter than previous bounds in the literature. In particular, the fast-rate bound is equivalent to the Seeger-Langford bound. Secondly, for losses with more general tail behaviors, we introduce two new parameter-free bounds: a PAC-Bayes Chernoff analogue when the loss' cumulative generating function is bounded, and a bound when the loss' second moment is bounded. These two bounds are obtained using a new technique based on a discretization of the space of possible events for the "in probability" parameter optimization problem. This technique is both simpler and more general than previous approaches optimizing over a grid on the parameters' space. Finally, using a simple technique that is applicable to any existing bound, we extend all previous results to anytime-valid bounds.

We introduce a new class of codebook-aware jamming strategies against coded transmissions over AWGN channels. The proposed strategies derive attack vectors from non-zero positions of minimum-weight codewords in Hamming space and utilize the geometry of minimum-distance error events in Euclidean space in their attack. We characterize the success probability of the attacker analytically and utilize these results for attack optimization. We demonstrate that the proposed jamming attacks are highly efficient; compared to Gaussian attack vectors with the same energy budget, the attacker's success probability is increased by more than two orders of magnitude while only consuming a fraction of the energy of the attacked codeword.

To date, no “information-theoretic” frameworks for reasoning about generalization error have been shown to establish minimax rates for gradient descent in the setting of stochastic convex optimization. In this work, we consider the prospect of establishing such rates via several existing information-theoretic frameworks: input-output mutual information bounds, conditional mutual information bounds and variants, PAC-Bayes bounds, and recent conditional variants thereof. We prove that none of these bounds are able to establish minimax rates. We then consider a common tactic employed in studying gradient methods, whereby the final iterate is corrupted by Gaussian noise, producing a noisy “surrogate” algorithm. We prove that minimax rates cannot be established via the analysis of such surrogates. Our results suggest that new ideas are required to analyze gradient descent using information-theoretic techniques. 

Background Whether early fluid accumulation is a risk factor for adverse renal outcomes in septic intensive care unit (ICU) patients remains uncertain. We assessed the association between cumulative fluid balance and major adverse kidney events within 30 days (MAKE30), a composite of death, dialysis, or sustained renal dysfunction, in such patients. Methods We performed a multicenter, retrospective observational study in 1834 septic patients admitted to five ICUs in three hospitals in Stockholm, Sweden. We used logistic regression analysis to assess the association between cumulative fluid balance during the first two days in ICU and subsequent risk of MAKE30, adjusted for demographic factors, comorbidities, baseline creatinine, illness severity variables, haemodynamic characteristics, chloride exposure and nephrotoxic drug exposure. We assessed the strength of significant exposure variables using a relative importance analysis. Results Overall, 519 (28.3%) patients developed MAKE30. Median (IQR) cumulative fluid balance was 5.3 (2.8-8.1) l in the MAKE30 group and 4.1 (1.9-6.8) l in the no MAKE30 group, with non-resuscitation fluids contributing to approximately half of total fluid input in each group. The adjusted odds ratio for MAKE30 was 1.05 (95% CI 1.02-1.09) per litre cumulative fluid balance. On relative importance analysis, the strongest factors regarding MAKE30 were, in decreasing order, baseline creatinine, cumulative fluid balance, and age. In the secondary outcome analysis, the adjusted odds ratio for dialysis or sustained renal dysfunction was 1.06 (95% CI 1.01-1.11) per litre cumulative fluid balance. On separate sensitivity analyses, lower urine output and early acute kidney injury, respectively, were independently associated with MAKE30, whereas higher fluid input was not. Conclusions In ICU patients with sepsis, a higher cumulative fluid balance after 2 days in ICU was associated with subsequent development of major adverse kidney events within 30 days, including death, renal replacement requirement, or persistent renal dysfunction.

In this article, we present a novel application of glide symmetry to differential lines with common-mode (CM) rejection filter properties. Two different topologies are investigated. First, glide symmetry is applied to a pair of differential lines where ground-connected mushrooms are employed as a CM rejection structure. The same idea is also used in a pair of differential lines where defective ground structures are introduced to stop the CM propagation. It is demonstrated that the CM rejection bandwidth is drastically increased when glide symmetry is exploited in both topologies when compared with their corresponding structures without glide symmetry. Furthermore, we show that the differential-mode propagation is hardly affected by the use of glide symmetry, ensuring the good integrity of the transmitted information. Experimental demonstration for both mushroom and defected ground structure is provided. Good agreement between simulations and measurements results is observed.

Power consumption is a major challenge for massive deployment of wireless sensors in internet of things (IoT) networks. This paper studies the use of analog joint source-channel (AJSCC) mappings in low-power sensing schemes. In particular, we propose a novel triangular mapping geometry as a low-complexity dimension reduction mapping. The proposed triangular mapping is employed for analog compression of multiple sensor readings into one signal, and thus, limits the need for power-hungry analog-to-digital conversion and processing at the sensing nodes. A comprehensive performance analysis of the proposed triangular mapping in terms of the mean squared error (MSE) performance is provided analytically and verified numerically. The problem of mapping adaptation to different source distributions is also studied. Moreover, the proposed triangular mapping is adopted in an energy scheduling problem in which the sensing nodes schedule their use of the received powers at different time instants and adjust the mapping parameters accordingly with the goal of minimizing the sum distortion at the receiver. We present a fast low-complexity algorithm for optimal energy scheduling and verify its performance in comparison with commercial convex optimization solvers. It is shown that the proposed mapping provides a very good MSE performance compared to the AJSCC benchmarks despite having a much lower complexity circuit implementation.