Nura Aljaafari

Nura Aljaafari, Marco Valentino, André Freitas

Predicting a label correctly does not necessarily require representing the operation that produces it. Transformer representations are known to carry label-level information, but whether they encode semantic operations producing those labels is unclear. We investigate this in Natural Language Inference using controlled premise-hypothesis pairs that differ by a single semantic transformation. Using layer-wise activations, we estimate operation-level subspaces via SVD and test their causal relevance through activation steering in four open-weight decoder models. Transformation effects are decodable with $84.8$-$99\%$ accuracy and occupy partially distinct but overlapping subspaces, exceeding random-subspace baselines. Steering experiments show that these directions causally influence predictions, though steerability varies across models; cross-operation steering further reveals structured interference and a dissociation between subspace selectivity and cross-operation independence. These findings indicate that the models encode not only that a hypothesis relates to a premise but also, in part, how it does so, implying that mechanistic analysis and control should operate at the level of semantic operations rather than predicted labels alone.

Mahmoud Nazzal, Nura Aljaafari, Ahmed Sawalmeh et al.

Federated learning enables multiple clients to collaboratively contribute to the learning of a global model orchestrated by a central server. This learning scheme promotes clients' data privacy and requires reduced communication overheads. In an application like network traffic classification, this helps hide the network vulnerabilities and weakness points. However, federated learning is susceptible to backdoor attacks, in which adversaries inject manipulated model updates into the global model. These updates inject a salient functionality in the global model that can be launched with specific input patterns. Nonetheless, the vulnerability of network traffic classification models based on federated learning to these attacks remains unexplored. In this paper, we propose GABAttack, a novel genetic algorithm-based backdoor attack against federated learning for network traffic classification. GABAttack utilizes a genetic algorithm to optimize the values and locations of backdoor trigger patterns, ensuring a better fit with the input and the model. This input-tailored dynamic attack is promising for improved attack evasiveness while being effective. Extensive experiments conducted over real-world network datasets validate the success of the proposed GABAttack in various situations while maintaining almost invisible activity. This research serves as an alarming call for network security experts and practitioners to develop robust defense measures against such attacks.

Nura Aljaafari, Danilo S. Carvalho, Andre Freitas

Mechanistic interpretability produces circuit-level causal analyses of neural network behaviour, but discovered circuits often remain isolated experimental artefacts: there is no shared formal representation for what circuits compute, how they relate, or when two findings provide evidence for the same mechanism. This work provides a formal infrastructure for cumulative mechanistic science by treating circuit interpretation as inductive theory construction. Each circuit is characterised at two levels: a Causal Functional Signature (CFS), which grounds component behaviour in causal attribution evidence and token role profiles, and an architectural signature $τ_{\mathrm{arch}}$, learned by inductive logic programming (ILP) from scale-invariant structural predicates. Together, these constitute a formal coherence layer that makes mechanistic claims explicit, comparable via $θ$-subsumption, and portable across model scales. CFS reveals qualitatively distinct computational strategies across task types, including attention-mediated copying versus MLP-mediated binding. ILP signatures achieve substantially better structural separation than graph kernel and feature-vector baselines, and support principled transfer across model scales and architecture families.

Nura Aljaafari, Danilo S. Carvalho, André Freitas

Locating and editing knowledge in large language models (LLMs) is crucial for enhancing their accuracy, safety, and inference rationale. We introduce ``concept editing'', an innovative variation of knowledge editing that uncovers conceptualisation mechanisms within these models. Using the reverse dictionary task, inference tracing, and input abstraction, we analyse the Multi-Layer Perceptron (MLP), Multi-Head Attention (MHA), and hidden state components of transformer models. Our results reveal distinct patterns: MLP layers employ key-value retrieval mechanism and context-dependent processing, which are highly associated with relative input tokens. MHA layers demonstrate a distributed nature with significant higher-level activations, suggesting sophisticated semantic integration. Hidden states emphasise the importance of the last token and top layers in the inference process. We observe evidence of gradual information building and distributed representation. These observations elucidate how transformer models process semantic information, paving the way for targeted interventions and improved interpretability techniques. Our work highlights the complex, layered nature of semantic processing in LLMs and the challenges of isolating and modifying specific concepts within these models.

Nura Aljaafari, Danilo S. Carvalho, André Freitas

Understanding the internal mechanisms of large language models (LLMs) is integral to enhancing their reliability, interpretability, and inference processes. We present Constituent-Aware Pooling (CAP), a methodology designed to analyse how LLMs process compositional linguistic structures. Grounded in principles of compositionality, mechanistic interpretability, and information theory, CAP systematically intervenes in model activations through constituent-based pooling at various model levels. Our experiments on inverse definition modelling, hypernym and synonym prediction reveal critical insights into transformers' limitations in handling compositional abstractions. No specific layer integrates tokens into unified semantic representations based on their constituent parts. We observe fragmented information processing, which intensifies with model size, suggesting that larger models struggle more with these interventions and exhibit greater information dispersion. This fragmentation likely stems from transformers' training objectives and architectural design, preventing systematic and cohesive representations. Our findings highlight fundamental limitations in current transformer architectures regarding compositional semantics processing and model interpretability, underscoring the critical need for novel approaches in LLM design to address these challenges.

Nura Aljaafari, Danilo S. Carvalho, André Freitas

Large language models (LLMs) struggle with compositional generalisation, limiting their ability to systematically combine learned components to interpret novel inputs. While architectural modifications, fine-tuning, and data augmentation improve compositionality, they often have limited adaptability, face scalability constraints, or yield diminishing returns on real data. To address this, we propose CARMA, an intervention that enhances the stability and robustness of compositional reasoning in LLMs while preserving fine-tuned performance. CARMA employs mutual information regularisation and layer-wise stability constraints to mitigate feature fragmentation, ensuring structured representations persist across and within layers. We evaluate CARMA on inverse dictionary modelling and sentiment classification, measuring its impact on semantic consistency, performance stability, and robustness to lexical perturbations. Results show that CARMA reduces the variability introduced by fine-tuning, stabilises token representations, and improves compositional reasoning. While its effectiveness varies across architectures, CARMA's key strength lies in reinforcing learned structures rather than introducing new capabilities, making it a scalable auxiliary method. These findings suggest that integrating CARMA with fine-tuning can improve compositional generalisation while maintaining task-specific performance in LLMs.

Nura Aljaafari, Danilo S. Carvalho, André Freitas

Despite displaying semantic competence, large language models' internal mechanisms that ground abstract semantic structure remain insufficiently characterised. We propose a method integrating role-cross minimal pairs, temporal emergence analysis, and cross-model comparison to study how LLMs implement semantic roles. Our analysis uncovers: (i) highly concentrated circuits (89-94% attribution within 28 nodes); (ii) gradual structural refinement rather than phase transitions, with larger models sometimes bypassing localised circuits; and (iii) moderate cross-scale conservation (24-59% component overlap) alongside high spectral similarity. These findings suggest that LLMs form compact, causally isolated mechanisms for abstract semantic structure, and these mechanisms exhibit partial transfer across scales and architectures.

Nura Aljaafari, Danilo S. Carvalho, André Freitas

Understanding when and how linguistic knowledge emerges during language model training remains a central challenge for interpretability. Most existing tools are post hoc, rely on scalar metrics, or require nontrivial integration effort, making comprehensive interpretability analysis difficult to deploy and maintain. We introduce TRACE, a modular toolkit for training and inference-time interpretability analysis of transformer models. It enables lightweight, in-training analysis of linguistic and representational signals, including features probing, intrinsic dimensionality, Hessian curvature, and output diagnostics. It integrates with ABSynth, a controllable synthetic corpus generator that provides structured annotations for precise evaluation of linguistic feature acquisition. Experiments with autoregressive transformers demonstrate that TRACE reveals developmental phenomena such as early syntactic emergence, delayed semantic acquisition, and representational compression, signals overlooked by traditional scalar metrics such as loss or accuracy. With minimal integration effort, the tool enables layer-wise diagnostics, convergence-based early stopping, and detection of structural errors, making transformer analysis interpretable, actionable, and reproducible.

Nura Aljaafari, Danilo S. Carvalho, André Freitas

Modern transformer models exhibit phase transitions during training, distinct shifts from memorisation to abstraction, but the mechanisms underlying these transitions remain poorly understood. Prior work has often focused on endpoint representations or isolated signals like curvature or mutual information, typically in symbolic or arithmetic domains, overlooking the emergence of linguistic structure. We introduce TRACE (Tracking Representation Abstraction and Compositional Emergence), a diagnostic framework combining geometric, informational, and linguistic signals to detect phase transitions in Transformer-based LMs. TRACE leverages a frame-semantic data generation method, ABSynth, that produces annotated synthetic corpora with controllable complexity, lexical distributions, and structural entropy, while being fully annotated with linguistic categories, enabling precise analysis of abstraction emergence. Experiments reveal that (i) phase transitions align with clear intersections between curvature collapse and dimension stabilisation; (ii) these geometric shifts coincide with emerging syntactic and semantic accuracy; (iii) abstraction patterns persist across architectural variants, with components like feedforward networks affecting optimisation stability rather than fundamentally altering trajectories. This work advances our understanding of how linguistic abstractions emerge in LMs, offering insights into model interpretability, training efficiency, and compositional generalisation that could inform more principled approaches to LM development.