Given a graph G and a positive integer k, the Graphlet Sampling problem asks to sample a connected induced k-vertex subgraph of G uniformly at random. Graphlet sampling enhances machine learning applications by transforming graph structures into feature vectors for tasks such as graph classification and subgraph identification, boosting neural network performance, and supporting clustered federated learning by capturing local structures and relationships.

Learning the unknown causal parameters of a linear structural causal model is a fundamental task in causal analysis. The task, known as the problem of identification, asks to estimate the parameters of the model from a combination of assumptions on the graphical structure of the model and observational data, represented as a non-causal covariance matrix. In this paper, we give a new sound and complete algorithm for generic identification which runs in polynomial space. By a standard simulation result, namely PSPACE EXP, this algorithm has exponential running time which vastly improves the state-of-the-art double exponential time method using a Gröbner basis approach. The paper also presents evidence that parameter identification is computationally hard in general. In particular, we prove, that the task asking whether, for a given feasible correlation matrix, there are exactly one or two or more parameter sets explaining the observed matrix, is hard for R, the co-class of the existential theory of the reals. In particular, this problem is coNP-hard. To our best knowledge, this is the first hardness result for some notion of identifiability.

In developing efficient optimization algorithms, it is crucial to account for communication constraints--a significant challenge in modern Federated Learning. The best-known communication complexity among non-accelerated algorithms is achieved by DANE, a distributed proximal-point algorithm that solves local subproblems at each iteration and that can exploit second-order similarity among individual functions. However, to achieve such communication efficiency, the algorithm requires solving local subproblems sufficiently accurately resulting in slightly sub-optimal local complexity. Inspired by the hybrid-projection proximalpoint method, in this work, we propose a novel distributed algorithm S-DANE. Compared to DANE, this method uses an auxiliary sequence of prox-centers while maintaining the same deterministic communication complexity. Moreover, the accuracy condition for solving the subproblem is milder, leading to enhanced local computation efficiency. Furthermore, S-DANE supports partial client participation and arbitrary stochastic local solvers, making it attractive in practice. We further accelerate S-DANE and show that the resulting algorithm achieves the best-known communication complexity among all existing methods for distributed convex optimization while still enjoying good local computation efficiency as S-DANE. Finally, we propose adaptive variants of both methods using line search, obtaining the first provably efficient adaptive algorithms that could exploit local second-order similarity without the prior knowledge of any parameters.

Reward functions are central in reinforcement learning (RL), guiding agents towards optimal decision-making. The complexity of RL tasks requires meticulously designed reward functions that effectively drive learning while avoiding unintended consequences. Effective reward design aims to provide signals that accelerate the agent's convergence to optimal behavior. Crafting rewards that align with task objectives, foster desired behaviors, and prevent undesirable actions is inherently challenging. This thesis delves into the critical role of reward signals in RL, highlighting their impact on the agent's behavior and learning dynamics and addressing challenges such as delayed, ambiguous, or intricate rewards. In this thesis work, we tackle different aspects of reward shaping. First, we address the problem of designing informative and interpretable reward signals from a teacher's/expert's perspective (teacher-driven). Here, the expert, equipped with the optimal policy and the corresponding value function, designs reward signals that expedite the agent's convergence to optimal behavior. Second, we build on this teacher-driven approach by introducing a novel method for adaptive interpretable reward design. In this scenario, the expert tailors the rewards based on the learner's current policy, ensuring alignment and optimal progression. Third, we propose a meta-learning approach, enabling the agent to self-design its reward signals online without expert input (agent-driven). This self-driven method considers the agent's learning and exploration to establish a self-improving feedback loop.

B-cos Networks have been shown to be effective for obtaining highly human interpretable explanations of model decisions by architecturally enforcing stronger alignment between inputs and weight. B-cos variants of convolutional networks (CNNs) and vision transformers (ViTs), which primarily replace linear layers with B-cos transformations, perform competitively to their respective standard variants while also yielding explanations that are faithful by design. However, it has so far been necessary to train these models from scratch, which is increasingly infeasible in the era of large, pre-trained foundation models. In this work, inspired by the architectural similarities in standard DNNs and B-cos networks, we propose'Bcosification', a novel approach to transform existing pre-trained models to become inherently interpretable. We perform a thorough study of design choices to perform this conversion, both for convolutional neural networks and vision transformers. We find that B-cosification can yield models that are on par with B-cos models trained from scratch in terms of interpretability, while often outperforming them in terms of classification performance at a fraction of the training cost. Subsequently, we apply B-cosification to a pretrained CLIP model, and show that, even with limited data and compute cost, we obtain a B-cosified version that is highly interpretable and competitive on zero shot performance across a variety of datasets.

Lipschitz gradient, without needing to know neither the corresponding Lipschitz constants, nor the oracle's variance but enjoying the rates which are characteristic for algorithms which have the

There has been enormous progress in the field of machine learning in the last decade, which has affected a variety of other areas as well. One of these areas is the design of online algorithms. Traditionally, the analysis of such algorithms involves worst-case guarantees, which can often be quite pessimistic. It is conceivable though, that having prior knowledge regarding the online input (obtained using machine learning) could potentially improve those guarantees significantly. In this work, we consider various online selection algorithms augmented with so-called machine learned advice. In particular, we consider secretary and online bipartite matching problems. The high-level idea is to incorporate some form of predictions in an existing online algorithm in order to get the best of two worlds: (i) provably improve the algorithm's performance guarantee in the case that predictions are sufficiently good, while (ii) losing only a constant factor of the algorithm's existing Work done in part while the author was at Saarland University and Max-Planck-Institute for Informatics and supported by DFG grant AN 1262/1-1.

After more than two decades of extensive study on the complexity of single-winner voting problems, the computational social choice community has recently shifted its primary focus to multiwinner voting, given its generality and broad applications. In particular, many variants of manipulation, control, and bribery problems for approval-based multiwinner voting rules (ABM rules for short) have been studied from a complexity point of view (see e.g., [2, 27, 48, 55]). Existing works in this line of research predominantly concern the constructive model of these problems, which models scenarios where a strategic agent attempts to elevate a single distinguished candidate to winner status, or make a committee a winning committee. However, the destructive counterparts of these problems have not been adequately studied in the literature so far. This paper studies the complexity and the parameterized complexity of several destructive bribery problems for ABM rules. These problems are designed to capture scenarios where an election attacker (or briber) aims to prevent multiple distinguished candidates from winning by making changes to the votes (e.g., by bribing some voters to alter their votes) under certain budget constraints. The attacker's motivation may stem from these distinguished candidates being rivals (e.g., having completely different political views from the attacker), or the attacker attempting to make them lose to increase the winning chance of her preferred candidates. We consider five bribery operations categorized into two classes: atomic operations and vote-level operations.

In machine learning (ML), it is common to account for multiple objectives when, e.g., selecting a model to deploy. However, it is often unclear how one should compare, aggregate and, ultimately, trade-off these objectives, as they might be measured in different units or scales. For example, when deploying large language models (LLMs), we might not only care about their performance, but also their CO2 consumption. In this work, we investigate how objectives can be sensibly compared and aggregated to navigate their Pareto front. To do so, we propose to make incomparable objectives comparable via their CDFs, approximated by their relative rankings. This allows us to aggregate them while matching user-specific preferences, allowing practitioners to meaningfully navigate and search for models in the Pareto front. We demonstrate the potential impact of our methodology in diverse areas such as LLM selection, domain generalization, and AutoML benchmarking, where classical ways to aggregate and normalize objectives fail.

Large language models (LLM) have revolutionized the processing of natural language. Although first benchmarks of the process modeling abilities of LLM are promising, it is currently under debate to what extent an LLM can generate good process models. In this contribution, we argue that the evaluation of the process modeling abilities of LLM is far from being trivial. Hence, available evaluation results must be taken carefully. For example, even in a simple scenario, not only the quality of a model should be taken into account, but also the costs and time needed for generation. Thus, an LLM does not generate one optimal solution, but a set of Pareto-optimal variants. Moreover, there are several further challenges which have to be taken into account, e.g. conceptualization of quality, validation of results, generalizability, and data leakage. We discuss these challenges in detail and discuss future experiments to tackle these challenges scientifically.