Large language models (LLMs) have been shown to possess impressive capabilities, while also raising crucial concerns about the faithfulness of their responses. A primary issue arising in this context is the management of (un)answerable queries by LLMs, which often results in hallucinatory behavior due to overconfidence. In this paper, we explore the behavior of LLMs when presented with (un)answerable queries. We ask: do models represent the fact that the question is (un)answerable when generating a hallucinatory answer? Our results show strong indications that such models encode the answerability of an input query, with the representation of the first decoded token often being a strong indicator. These findings shed new light on the spatial organization within the latent representations of LLMs, unveiling previously unexplored facets of these models. Moreover, they pave the way for the development of improved decoding techniques with better adherence to factual generation, particularly in scenarios where query (un)answerability is a concern.

The recently introduced Controlled Text Reduction (CTR) task isolates the text generation step within typical summarization-style tasks. It does so by challenging models to generate coherent text conforming to pre-selected content within the input text (``highlights''). This framing enables increased modularity in summarization-like tasks, allowing to couple a single CTR model with various content-selection setups and modules. However, there are currently no reliable CTR models, while the performance of the existing baseline for the task is mediocre, falling short of practical utility. Here, we address this gap by introducing a high-quality, open-source CTR model that tackles two prior key limitations: inadequate enforcement of the content-preservation constraint, and suboptimal silver training data. Addressing these, we amplify the content-preservation constraint in both training, via RL, and inference, via a controlled decoding strategy. Further, we substantially improve the silver training data quality via GPT-4 distillation. Overall, pairing the distilled dataset with the highlight-adherence strategies yields marked gains over the current baseline, of up to 30 ROUGE-L points, providing a reliable CTR model for downstream use.

Fusing multiple modalities has proven effective for multimodal information processing. However, the incongruity between modalities poses a challenge for multimodal fusion, especially in affect recognition. In this study, we first analyze how the salient affective information in one modality can be affected by the other, and demonstrate that inter-modal incongruity exists latently in crossmodal attention. Based on this finding, we propose the Hierarchical Crossmodal Transformer with Dynamic Modality Gating (HCT-DMG), a lightweight incongruity-aware model, which dynamically chooses the primary modality in each training batch and reduces fusion times by leveraging the learned hierarchy in the latent space to alleviate incongruity. The experimental evaluation on five benchmark datasets: CMU-MOSI, CMU-MOSEI, and IEMOCAP (sentiment and emotion), where incongruity implicitly lies in hard samples, as well as UR-FUNNY (humour) and MUStaRD (sarcasm), where incongruity is common, verifies the efficacy of our approach, showing that HCT-DMG: 1) outperforms previous multimodal models with a reduced size of approximately 0.8M parameters; 2) recognizes hard samples where incongruity makes affect recognition difficult; 3) mitigates the incongruity at the latent level in crossmodal attention.

Examples of bias and discrimination in AI applications include court decisions [1], job hiring [2], online ads [3], and many other areas prone to bias [4]. These algorithmic decisions have economic and personal implications for individuals. Therefore, Fairness, Accountability, Transparency and Ethics (FATE) in AI must be properly regulated for responsible use cases [5, 6], particularly in high-stakes domains [1, 7, 8, 9, 10, 11, 12]. Studies have shown that machine learning models can discriminate based on race and gender [13, 14, 15]. FATE in AI is intended to address the social issues caused by digital systems, but the current discourse is largely shaped by more economically developed countries (MEDC), raising concerns about neglecting local knowledge, cultural pluralism, and global fairness [16]. As AI systems become more integrated into various products [9, 10, 17, 12, 18, 19], they are a major driver of the fourth industrial revolution (4IR) and transformation [20]. Therefore, it is essential to understand the FATE-related needs of different communities, as AI affects a wide range of people. Ensuring effective transparency cannot be a one-size-fits-all approach [21], as this could disproportionately affect different communities [16, 22]. To this end, more contextualised and interdisciplinary research is needed to inform algorithmic fairness and transparency [23, 24, 25].

Three challenges, however, can hinder the application of Feedback Linearization: over-intensive control signals, singular decoupling matrix, and saturation. Activating any of these three issues can challenge the stability proof. To solve these three challenges, first, this research proposed the drone gait plan. The gait plan was initially used to figure out the control problems in quadruped (four-legged) robots; applying this approach, accompanied by Feedback Linearization, the quality of the control signals was enhanced. Then, we proposed the concept of unacceptable attitude curves, which are not allowed for the tiltrotor to travel to. The Two Color Map Theorem was subsequently established to enlarge the supported attitude for the tiltrotor. These theories were employed in the tiltrotor tracking problem with different references. Notable improvements in the control signals were witnessed in the tiltrotor simulator. Finally, we explored the control theory, the stability proof of the novel mobile robot (tilt vehicle) stabilized by Feedback Linearization with saturation. Instead of adopting the tiltrotor model, which is over-complicated, we designed a conceptual mobile robot (tilt-car) to analyze the stability proof. The stability proof (stable in the sense of Lyapunov) was found for a mobile robot (tilt vehicle) controlled by Feedback Linearization with saturation for the first time. The success tracking result with the promising control signals in the tiltrotor simulator demonstrates the advances of our control method. Also, the Lyapunov candidate and the tracking result in the mobile robot (tilt-car) simulator confirm our deductions of the stability proof. These results reveal that these three challenges in Feedback Linearization are solved, to some extents.

We consider the learning dynamics of a single reinforcement learning optimal execution trading agent when it interacts with an event driven agent-based financial market model. Trading takes place asynchronously through a matching engine in event time. The optimal execution agent is considered at different levels of initial order-sizes and differently sized state spaces. The resulting impact on the agent-based model and market are considered using a calibration approach that explores changes in the empirical stylised facts and price impact curves. Convergence, volume trajectory and action trace plots are used to visualise the learning dynamics. Here the smaller state space agents had the number of states they visited converge much faster than the larger state space agents, and they were able to start learning to trade intuitively using the spread and volume states. We find that the moments of the model are robust to the impact of the learning agents except for the Hurst exponent, which was lowered by the introduction of strategic order-splitting. The introduction of the learning agent preserves the shape of the price impact curves but can reduce the trade-sign auto-correlations when their trading volumes increase.

It is commonly known that the conjunctive query entailment problem for certain extensions of (the well-known ontology language) ALC is computationally harder than their knowledge base satisfiability problem while for others the complexities coincide, both under the standard and the finite-model semantics. We expose a uniform principle behind this divide by identifying a wide class of (finitely) locally-forward description logics, for which we prove that (finite) query entailment problem can be solved by a reduction to exponentially many calls of the (finite) knowledge base satisfiability problem. Consequently, our algorithm yields tight ExpTime upper bounds for locally-forward logics with ExpTime-complete knowledge base satisfiability problem, including logics between ALC and µALCHbregQ (and more), as well as ALCSCC with global cardinality constraints, for which the complexity of querying remained open. Moreover, to make our technique applicable in future research, we provide easy-to-check sufficient conditions for a logic to be locally-forward based several versions of the on model-theoretic notion of unravellings. Together with existing results, this provides a nearly complete classification of the “benign” vs. “malign” primitive modelling features extending ALC, missing out only the Self operator. We then show a rather counter-intuitive result, namely that the conjunctive entailment problem for ALCSelf is exponentially harder than for ALC. This places the seemingly innocuous Self operator among the “malign” modelling features, like inverses, transitivity or nominals.

Categorizing source codes accurately and efficiently is a challenging problem in real-world programming education platform management. In recent years, model-based approaches utilizing abstract syntax trees (ASTs) have been widely applied to code classification tasks. We introduce an approach named the Sparse Attention-based neural network for Code Classification (SACC) in this paper. The approach involves two main steps: In the first step, source code undergoes syntax parsing and preprocessing. The generated abstract syntax tree is split into sequences of subtrees and then encoded using a recursive neural network to obtain a high-dimensional representation. This step simultaneously considers both the logical structure and lexical level information contained within the code. In the second step, the encoded sequences of subtrees are fed into a Transformer model that incorporates sparse attention mechanisms for the purpose of classification. This method efficiently reduces the computational cost of the self-attention mechanisms, thus improving the training speed while preserving effectiveness. Our work introduces a carefully designed sparse attention pattern that is specifically designed to meet the unique needs of code classification tasks. This design helps reduce the influence of redundant information and enhances the overall performance of the model. Finally, we also deal with problems in previous related research, which include issues like incomplete classification labels and a small dataset size. We annotated the CodeNet dataset with algorithm-related labeling categories, which contains a significantly large amount of data. Extensive comparative experimental results demonstrate the effectiveness and efficiency of SACC for the code classification tasks.

This paper studies the problem of modeling interacting dynamical systems, which is critical for understanding physical dynamics and biological processes. Recent research predominantly uses geometric graphs to represent these interactions, which are then captured by powerful graph neural networks (GNNs). However, predicting interacting dynamics in challenging scenarios such as out-of-distribution shift and complicated underlying rules remains unsolved. In this paper, we propose a new approach named Graph ODE with factorized prototypes (GOAT) to address the problem. The core of GOAT is to incorporate factorized prototypes from contextual knowledge into a continuous graph ODE framework. Specifically, GOAT employs representation disentanglement and system parameters to extract both object-level and system-level contexts from historical trajectories, which allows us to explicitly model their independent influence and thus enhances the generalization capability under system changes. Then, we integrate these disentangled latent representations into a graph ODE model, which determines a combination of various interacting prototypes for enhanced model expressivity. The entire model is optimized using an end-to-end variational inference framework to maximize the likelihood. Extensive experiments in both in-distribution and out-of-distribution settings validate the superiority of GOAT.

Added toxicity in the context of translation refers to the fact of producing a translation output with more toxicity than there exists in the input. In this paper, we present MinTox which is a novel pipeline to identify added toxicity and mitigate this issue which works at inference time. MinTox uses a toxicity detection classifier which is multimodal (speech and text) and works in languages at scale. The mitigation method is applied to languages at scale and directly in text outputs. MinTox is applied to SEAMLESSM4T, which is the latest multimodal and massively multilingual machine translation system. For this system, MinTox achieves significant added toxicity mitigation across domains, modalities and language directions. MinTox manages to approximately filter out from 25% to 95% of added toxicity (depending on the modality and domain) while keeping translation quality.