Joanna Sliwa

Joanna Sliwa, Shubhangi Ghosh, Vincent Stimper et al.

One aim of representation learning is to recover the original latent code that generated the data, a task which requires additional information or inductive biases. A recently proposed approach termed Independent Mechanism Analysis (IMA) postulates that each latent source should influence the observed mixtures independently, complementing standard nonlinear independent component analysis, and taking inspiration from the principle of independent causal mechanisms. While it was shown in theory and experiments that IMA helps recovering the true latents, the method's performance was so far only characterized when the modeling assumptions are exactly satisfied. Here, we test the method's robustness to violations of the underlying assumptions. We find that the benefits of IMA-based regularization for recovering the true sources extend to mixing functions with various degrees of violation of the IMA principle, while standard regularizers do not provide the same merits. Moreover, we show that unregularized maximum likelihood recovers mixing functions which systematically deviate from the IMA principle, and provide an argument elucidating the benefits of IMA-based regularization.

Joanna Sliwa, Frank Schneider, Philipp Hennig et al.

Parameter-efficient fine-tuning methods, such as Low-Rank Adaptation (LoRA), enable fast specialization of large pre-trained models to different downstream applications. However, this process often leads to catastrophic forgetting of the model's prior domain knowledge. We address this issue with LaLoRA, a weight-space regularization technique that applies a Laplace approximation to Low-Rank Adaptation. Our approach estimates the model's confidence in each parameter and constrains updates in high-curvature directions, preserving prior knowledge while enabling efficient target-domain learning. By applying the Laplace approximation only to the LoRA weights, the method remains lightweight. We evaluate LaLoRA by fine-tuning a Llama model for mathematical reasoning and demonstrate an improved learning-forgetting trade-off, which can be directly controlled via the method's regularization strength. We further explore different loss landscape curvature approximations for estimating parameter confidence, analyze the effect of the data used for the Laplace approximation, and study robustness across hyperparameters.

Alessandro Aloisio, Marco Autili, Alfredo D'Angelo et al.

Due to the increasing complexity and heterogeneity of contemporary Command, Control, Communications, Computers, & Intelligence systems at all levels within military organizations, the adoption of the Service Oriented Architectures (SOA) principles and concepts is becoming essential. SOA provides flexibility and interoperability of services enabling the realization of efficient and modular information infrastructure for command and control systems. However, within a tactical domain, the presence of potentially highly mobile actors equipped with constrained communications media (i.e., unreliable radio networks with limited bandwidth) limits the applicability of traditional SOA technologies. The TACTICS project aims at the definition and experimental demonstration of a Tactical Services Infrastructure enabling tactical radio networks (without any modifications of the radio part of those networks) to participate in SOA infrastructures and provide, as well as consume, services to and from the strategic domain independently of the user's location.