Modularity is a compelling solution to continual learning (CL), the problem of modeling sequences of related tasks.

Large language model (LLM) agents have demonstrated strong capabilities across diverse domains, yet automated agent design remains a significant challenge. Current automated agent design approaches are often constrained by limited search spaces that primarily optimize workflows but fail to integrate crucial human-designed components like memory, planning, and tool use. Furthermore, these methods are hampered by high evaluation costs, as evaluating even a single new agent on a benchmark can require tens of dollars. The difficulty of this exploration is further exacerbated by inefficient search strategies that struggle to navigate the large design space effectively, making the discovery of novel agents a slow and resource-intensive process. To address these challenges, we propose AgentSwift, a novel framework for automated agent design. We formalize a hierarchical search space that jointly models agentic workflow and composable functional components. This structure moves beyond optimizing workflows alone by co-optimizing functional components, which enables the discovery of more complex and effective agent architectures. To make exploration within this expansive space feasible, we mitigate high evaluation costs by training a value model on a high-quality dataset, generated via a novel strategy combining combinatorial coverage and balanced Bayesian sampling for low-cost evaluation. Guiding the entire process is a hierarchical MCTS strategy, which is informed by uncertainty to efficiently navigate the search space. Evaluated across a comprehensive set of seven benchmarks spanning embodied, math, web, tool, and game domains, AgentSwift discovers agents that achieve an average performance gain of 8.34\% over both existing automated agent search methods and manually designed agents. Our framework serves as a launchpad for researchers to rapidly discover powerful agent architectures.

Most explainable AI (XAI) frameworks are limited in their expressiveness, summarizing complex feature effects as single scalar values ϕ_i. This approach answers "what" features are important but fails to reveal "how" they interact. Furthermore, methods that attempt to capture interactions, like those based on Shapley values, often face an exponential computational cost. We present STRIDE, a scalable framework that addresses both limitations by reframing explanation as a subset-enumeration-free, orthogonal "functional decomposition" in a Reproducing Kernel Hilbert Space (RKHS). In the tabular setups we study, STRIDE analytically computes functional components f_S(x_S) via a recursive kernel-centering procedure. The approach is model-agnostic and theoretically grounded with results on orthogonality and L^2 convergence. In tabular benchmarks (10 datasets, median over 10 seeds), STRIDE attains a 3.0 times median speedup over TreeSHAP and a mean R^2=0.93 for reconstruction. We also introduce "component surgery", a diagnostic that isolates a learned interaction and quantifies its contribution; on California Housing, removing a single interaction reduces test R^2 from 0.019 to 0.027.

In complex autonomous driving (AD) software systems, the functioning of each system part is crucial for safe operation. By measuring the current functionality or operability of individual components an isolated glimpse into the system is given. Literature provides several of these detached assessments, often in the form of safety or performance measures. But dependencies, redundancies, error propagation and conflicting functionality statements do not allow for easy combination of these measures into a big picture of the functioning of the entire AD stack. Data is processed and exchanged between different components, each of which can fail, making an overall statement challenging. The lack of functionality assessment frameworks that tackle these problems underlines this complexity. This article presents a novel framework for inferring an overall functionality statement for complex component based systems by considering their dependencies, redundancies, error propagation paths and the assessments of individual components. Our framework first incorporates a comprehensive conversion to an assessment representation of the system. The representation is based on Subjective Networks (SNs) that allow for easy identification of faulty system parts. Second, the framework offers a flexible method for computing the system's functionality while dealing with contradicting assessments about the same component and dependencies, as well as redundancies, of the system. We discuss the framework's capabilities on real-life data of our AD stack with assessments of various components.

The 3D continuum presents a complex environment that spans the terrestrial, aerial and space domains, with 6Gnetworks serving as a key enabling technology. Current AI approaches for network management rely on monolithic models that fail to capture cross-domain interactions, lack adaptability,and demand prohibitive computational resources. This paper presents a formal model of Compound AI systems, introducing a novel tripartite framework that decomposes complex tasks into specialized, interoperable modules. The proposed modular architecture provides essential capabilities to address the unique challenges of 6G networks in the 3D continuum, where heterogeneous components require coordinated, yet distributed, intelligence. This approach introduces a fundamental trade-off between model and system performance, which must be carefully addressed. Furthermore, we identify key challenges faced by Compound AI systems within 6G networks operating in the 3D continuum, including cross-domain resource orchestration, adaptation to dynamic topologies, and the maintenance of consistent AI service quality across heterogeneous environments.

Pre-trained transformer large language models (LLMs) demonstrate strong knowledge recall capabilities. This paper investigates the knowledge recall mechanism in LLMs by abstracting it into a functional structure. We propose that during knowledge recall, the model's hidden activation space implicitly entails a function execution process where specific activation vectors align with functional components (Input argument, Function body, and Return values). Specifically, activation vectors of relation-related tokens define a mapping function from subjects to objects, with subject-related token activations serving as input arguments and object-related token activations as return values. For experimental verification, we first design a patching-based knowledge-scoring algorithm to identify knowledge-aware activation vectors as independent functional components. Then, we conduct counter-knowledge testing to examine the independent functional effects of each component on knowledge recall outcomes. From this functional perspective, we improve the contextual knowledge editing approach augmented by activation patching. By rewriting incoherent activations in context, we enable improved short-term memory retention for new knowledge prompting.

In this paper, we establish minimax optimal rates of convergence for prediction in a semi-functional linear model that consists of a functional component and a less smooth nonparametric component. Our results reveal that the smoother functional component can be learned with the minimax rate as if the nonparametric component were known. More specifically, a double-penalized least squares method is adopted to estimate both the functional and nonparametric components within the framework of reproducing kernel Hilbert spaces. By virtue of the representer theorem, an efficient algorithm that requires no iterations is proposed to solve the corresponding optimization problem, where the regularization parameters are selected by the generalized cross validation criterion. Numerical studies are provided to demonstrate the effectiveness of the method and to verify the theoretical analysis.

Learning to solve sequential tasks with recurrent models requires the ability to memorize long sequences and to extract task-relevant features from them. In this paper, we study the memorization subtask from the point of view of the design and training of recurrent neural networks. We propose a new model, the Linear Memory Network, which features an encoding-based memorization component built with a linear autoencoder for sequences. We extend the memorization component with a modular memory that encodes the hidden state sequence at different sampling frequencies. Additionally, we provide a specialized training algorithm that initializes the memory to efficiently encode the hidden activations of the network. The experimental results on synthetic and real-world datasets show that specializing the training algorithm to train the memorization component always improves the final performance whenever the memorization of long sequences is necessary to solve the problem.

Recurrent neural networks can learn complex transduction problems that require maintaining and actively exploiting a memory of their inputs. Such models traditionally consider memory and input-output functionalities indissolubly entangled. We introduce a novel recurrent architecture based on the conceptual separation between the functional input-output transformation and the memory mechanism, showing how they can be implemented through different neural components. By building on such conceptualization, we introduce the Linear Memory Network, a recurrent model comprising a feedforward neural network, realizing the non-linear functional transformation, and a linear autoencoder for sequences, implementing the memory component. The resulting architecture can be efficiently trained by building on closed-form solutions to linear optimization problems. Further, by exploiting equivalence results between feedforward and recurrent neural networks we devise a pretraining schema for the proposed architecture. Experiments on polyphonic music datasets show competitive results against gated recurrent networks and other state of the art models.

Numerous methods for probabilistic reasoning in large, complex belief or decision networks are currently being developed. There has been little research on automating the dynamic, incremental construction of decision models. A uniform value-driven method of decision model construction is proposed for the hierarchical complete diagnosis. Hierarchical complete diagnostic reasoning is formulated as a stochastic process and modeled using influence diagrams. Given observations, this method creates decision models in order to obtain the best actions sequentially for locating and repairing a fault at minimum cost. This method construct decision models incrementally, interleaving probe actions with model construction and evaluation. The method treats meta-level and baselevel tasks uniformly. That is, the method takes a decision-theoretic look at the control of search in causal pathways and structural hierarchies.