Search Results (10)

Recent language models exhibit significant logical reasoning abilities, yet the mechanisms supporting deductive inference remain poorly understood. This paper studies small transformer-based language models trained from scratch on multi-step deductive tasks, focusing on the distinction between horizontal reasoning, where intermediate steps are generated autoregressively, and vertical reasoning, where inference unfolds implicitly across layers before the first output token is produced. We analyze two synthetic tasks: logical consequence over chains of symbolic implications and root-to-leaf navigation in binary trees. Mechanistic interpretability reveals that Chain-of-Thought supervision enables models to learn rule-based inference rather than statistical shortcuts. In the horizontal setting, a shallow attention-only model develops interpretable circuits for rule completion, rule chaining, and final decision making, largely implemented through induction-head-like mechanisms. We further introduce a truncated pseudoinverse method to decode the information carried by queries, keys, and values. For vertical reasoning, Chain-of-Thought appears to act less as explicit step-by-step guidance and more as a form of curriculum learning, helping the model acquire increasingly complex reasoning patterns. Without Chain-of-Thought, models tend to memorize or exploit dataset biases. These results provide a low-level account of how transformers can implement deductive reasoning and suggest how Chain-of-Thought may serve different functions in horizontal and vertical reasoning. Full article

Whether learned cognition can affect evolutionary outcomes remains a long-standing question. This study addresses a narrower mechanism: whether a model-based planner benefits from learned rules that explicitly condition on the action just taken. The testbed is a spatial artificial ecology with plants, shelters, a predator, reproduction, and a day/night cycle. Five rule-use arms are evaluated on matched simulation seeds. At age 200, agents switch to a weaker learned-lite planner that relies more strongly on learned rule predictions. The pre-specified hypothesis is that access to filtered action-conditioned rules improves outcomes relative to an otherwise identical no-rule-policy baseline, in which rules are still induced and stored but are not used for action selection. In thirty paired replicates under the default reproductive gates, the action-conditioned arm outperforms the no-rule baseline on all four pre-specified primary endpoints. The strongest effect is behavioural: the action arm produces 91.4 additional successful post-switch eating events per run ($d_z=1.56$, 93.3% paired win rate, $p<{10}^{-4}$). It also produces 10 additional crystallized clean-causal rules per replicate ($d_z=0.58$, $p_t=0.0034$). All four primary paired-t p-values remain significant after Bonferroni correction across the four-endpoint family. A diagnostic check shows that omitting reproductive cooldown from the planner’s rollout reverses the arm ordering on the same paired seeds; reinstating cooldown recovers the reported result. Two exploratory checks delimit the claim: broad unfiltered rule access can impair foraging, and a means–ends extension shifts behaviour toward reproduction without producing a robust whole-life fitness gain. Within this simulation, access to action-conditioned rules has a measurable effect on post-switch behaviour that is distinct from passive environmental prediction and from clean-crystallized rules alone. Full article

Background: Breakpoints in the pre-examination processes and at organizational interfaces are a significant source of failures in specimen identification and tracking in anatomic pathology. While ISO 15189 emphasizes end-to-end traceability and risk-based quality management, implementing these principles in complex, multi-actor specimen pathways remains challenging. This study explores expert perspectives on specimen process chains, tracking mechanisms, and ISO 15189-oriented quality and risk management in pathology. Methods: We conducted 10 semi-structured expert interviews across three settings. Interviews were audio-recorded, transcribed, pseudonymized, and analyzed using structured qualitative content analysis (Mayring) supported by MAXQDA. A deductive category system derived from the theoretical framework and interview guide comprised six main categories and twelve subcategories. Results: Across 512 coded text segments, participants identified several factors as critical for effective implementation, including: (i) interface management along the specimen pathway, with recurrent vulnerabilities at handovers between operating theater/ward/transport and accessioning; (ii) the central role of barcode-based identification and the need for closed-loop traceability; (iii) the importance of measurable quality indicators and incident learning systems to operationalize risk management; (iv) persistent paper–digital handoffs and heterogeneous IT landscapes that undermine data integrity; (v) the need for clearly assigned responsibilities, training, and SOP governance; and (vi) implementation barriers including resources, change management, and vendor integration, alongside practical enablers such as incremental roll-out and cross-professional governance. Conclusions: Experts converge on a pragmatic ISO 15189-aligned roadmap: prioritize interface risks, standardize identifiers and handover rules, define a minimal KPI set for tracking and misidentification events, and reduce paper–digital handoffs by interoperable IT. Future work should quantify baseline error rates and evaluate the impact of digital tracking interventions on patient safety and turnaround times. Full article

With the widespread application of automatic fire alarm systems in various types of buildings, the problem of fire false alarms has gradually become prominent, which not only causes resource waste, but also may reduce users’ trust in the alarm system, thereby affecting the efficiency of emergency response in actual fires. According to data from a certain fire cloud platform, 99.85% of the suspected fires predicted by its system are false alarms. Although existing models can recognize most fire accidents, the accuracy of fire alarm recognition is only 0.15%, due to loose judgment logic, which still requires a large amount of manpower to verify alarms. This article analyzes a large amount of false alarm data and explores the main causes of false alarms, including environmental interference, equipment failure, and improper human operation. By using a fire dynamics simulator (FDS) to establish fire simulation models under different data settings, horizontal and vertical multi-scene fire simulation data are obtained. The study combines simulation and platform data to form a fire and false alarm dataset using a one-dimensional convolutional neural network (1D-CNN) and deep neural network (DNN) deep learning techniques to learn the deductive rules of the fire scene, establish a two-stage judgment model, and gradually, accurately, judge the results. By quantifying the precision, recall, and F1 score of the model, a deep learning model designed to accurately identify genuine fire alarms while filtering out false ones is proposed that can significantly reduce the false alarm rate. The results indicate that the model can identify 1705 false alarms out of 2255 highly suspected true alarms identified by existing systems in multiple practical scenarios and eliminate 75.61% of false positive alarms. On the premise of ensuring an authenticity recognition rate greater than 98%, the accuracy of fire alarm recognition increased from 0.15% to 28.85%, which will significantly reduce the workload of staff verifying alerts, and has good practical value. Full article

Maintaining logical consistency in structured explanations is critical for understanding and troubleshooting the reasoning behind a system’s decisions. However, existing methods for entailment tree generation often struggle with logical consistency, resulting in erroneous intermediate conclusions and reducing the overall accuracy of the explanations. To address this issue, we propose case-based deduction (CBD), a novel approach that retrieves cases with similar logical structures from a case base and uses them as demonstrations for logical deduction. This method guides the model toward logically sound conclusions without the need for manually constructing logical rule bases. By leveraging a prototypical network for case retrieval and reranking them using information entropy, CBD introduces diversity to improve in-context learning. Our experimental results on the EntailmentBank dataset show that CBD significantly improves entailment tree generation, achieving performance improvements of 1.7% in Task 1, 0.6% in Task 2, and 0.8% in Task 3 under the strictest Overall AllCorrect metric. These findings confirm that CBD enhances the logical consistency and overall accuracy of AI systems in structured explanation tasks. Full article

To enhance the anti-submarine and search capabilities of multiple Unmanned Aerial Vehicle (UAV) groups in complex marine environments, this paper proposes a flexible action-evaluation algorithm known as Knowledge-Driven Soft Actor-Critic (KD-SAC), which can effectively interact with real-time environmental information. KD-SAC is a reinforcement learning algorithm that consists of two main components: UAV Group Search Knowledge Base (UGSKB) and path planning strategy. Firstly, based on the UGSKB, we establish a cooperation search framework that comprises three layers of information models: the data layer provides prior information and fundamental search rules to the system, the knowledge layer enriches search rules and database in continuous searching processes, and the decision layer utilizes above two layers of information models to enable autonomous decision-making by UAVs. Secondly, we propose a rule-based deductive inference return visit (RDIRV) strategy to enhance the knowledge base of search. The core concept of this strategy is to enable UAVs to learn from both successful and unsuccessful experiences, thereby enriching the search rules based on optimal decisions as exemplary cases. This approach can significantly enhance the learning performance of KD-SAC. The subsequent step involves designing an event-based UGSKB calling mechanism at the decision-making level, which calls a template based on the target and current motion. Finally, it uses a punishment function, and is then employed to achieve optimal decision-making for UAV actions and states. The feasibility and superiority of our proposed algorithm are demonstrated through experimental comparisons with alternative methods. The final results demonstrate that the proposed method achieves a success rate of 73.63% in multi-UAV flight path planning within complex environments, surpassing the other three algorithms by 17.27%, 29.88%, and 33.51%, respectively. In addition, the KD-SAC algorithm outperforms the other three algorithms in terms of synergy and average search reward. Full article

Evolutionary optimization (EO) has been proven to be highly effective computation means in solving asymmetry problems in engineering practices. In this study, a novel evolutionary optimization approach for the belief rule base (BRB) system is proposed, namely EO-BRB, by constructing an optimization model and employing the Differential Evolutionary (DE) algorithm as its optimization engine due to its ability to locate an optimal solution for problems with nonlinear complexity. In the EO-BRB approach, the most representative referenced values of the attributes which are pre-determined in traditional learning approaches are to be optimized. In the optimization model, the mean squared error (MSE) between the actual and observed data is taken as the objective, while the initial weights of all the rules, the beliefs of the scales in the conclusion part, and the referenced values of the attributes are taken as the restraints. Compared with the traditional learning approaches for the BRB system, the EO-BRB approach (1) does not require transforming the numerical referenced values of the attributes into linguistic terms; (2) does not require identifying any initial solution; (3) does not require any mathematical deduction and/or case-specific information which verifies it as a general approach; and (4) can help downsize the BRB system while producing superior performances. Thus, the proposed EO-BRB approach can make the best use of the nonlinear modeling ability of BRB and the optimization superiority of the EO algorithms. Three asymmetry numerical and practical cases are studied to validate the efficiency of the proposed EO-BRB approach. Full article

The reinforcement learning problem of complex action control in a multi-player wargame has been a hot research topic in recent years. In this paper, a game system based on turn-based confrontation is designed and implemented with state-of-the-art deep reinforcement learning models. Specifically, we first design a Q-learning algorithm to achieve intelligent decision-making, which is based on the DQN (Deep Q Network) to model complex game behaviors. Then, an a priori knowledge-based algorithm PK-DQN (Prior Knowledge-Deep Q Network) is introduced to improve the DQN algorithm, which accelerates the convergence speed and stability of the algorithm. The experiments demonstrate the correctness of the PK-DQN algorithm, it is validated, and its performance surpasses the conventional DQN algorithm. Furthermore, the PK-DQN algorithm shows effectiveness in defeating the high level of rule-based opponents, which provides promising results for the exploration of the field of smart chess and intelligent game deduction. Full article

According to literature and despite their commercial success, state-of-the-art two-stage non-iterative geographic object-based image analysis (GEOBIA) systems and three-stage iterative geographic object-oriented image analysis (GEOOIA) systems, where GEOOIA/GEOBIA, remain affected by a lack of productivity, general consensus and research. To outperform the Quality Indexes of Operativeness (OQIs) of existing GEOBIA/GEOOIA systems in compliance with the Quality Assurance Framework for Earth Observation (QA4EO) guidelines, this methodological work is split into two parts. Based on an original multi-disciplinary Strengths, Weaknesses, Opportunities and Threats (SWOT) analysis of the GEOBIA/GEOOIA approaches, the first part of this work promotes a shift of learning paradigm in the pre-attentive vision first stage of a remote sensing (RS) image understanding system (RS-IUS), from sub-symbolic statistical model-based (inductive) image segmentation to symbolic physical model-based (deductive) image preliminary classification capable of accomplishing image sub-symbolic segmentation and image symbolic pre-classification simultaneously. In the present second part of this work, a novel hybrid (combined deductive and inductive) RS-IUS architecture featuring a symbolic deductive pre-attentive vision first stage is proposed and discussed in terms of: (a) computational theory (system design), (b) information/knowledge representation, (c) algorithm design and (d) implementation. As proof-of-concept of symbolic physical model-based pre-attentive vision first stage, the spectral knowledge-based, operational, near real-time, multi-sensor, multi-resolution, application-independent Satellite Image Automatic Mapper™ (SIAM™) is selected from existing literature. To the best of these authors’ knowledge, this is the first time a symbolic syntactic inference system, like SIAM™, is made available to the RS community for operational use in a RS-IUS pre-attentive vision first stage, to accomplish multi-scale image segmentation and multi-granularity image pre-classification simultaneously, automatically and in near real-time. Full article

According to existing literature and despite their commercial success, state-of-the-art two-stage non-iterative geographic object-based image analysis (GEOBIA) systems and three-stage iterative geographic object-oriented image analysis (GEOOIA) systems, where GEOOIA/GEOBIA, remain affected by a lack of productivity, general consensus and research. To outperform the degree of automation, accuracy, efficiency, robustness, scalability and timeliness of existing GEOBIA/GEOOIA systems in compliance with the Quality Assurance Framework for Earth Observation (QA4EO) guidelines, this methodological work is split into two parts. The present first paper provides a multi-disciplinary Strengths, Weaknesses, Opportunities and Threats (SWOT) analysis of the GEOBIA/GEOOIA approaches that augments similar analyses proposed in recent years. In line with constraints stemming from human vision, this SWOT analysis promotes a shift of learning paradigm in the pre-attentive vision first stage of a remote sensing (RS) image understanding system (RS-IUS), from sub-symbolic statistical model-based (inductive) image segmentation to symbolic physical model-based (deductive) image preliminary classification. Hence, a symbolic deductive pre-attentive vision first stage accomplishes image sub-symbolic segmentation and image symbolic pre-classification simultaneously. In the second part of this work a novel hybrid (combined deductive and inductive) RS-IUS architecture featuring a symbolic deductive pre-attentive vision first stage is proposed and discussed in terms of: (a) computational theory (system design); (b) information/knowledge representation; (c) algorithm design; and (d) implementation. As proof-of-concept of symbolic physical model-based pre-attentive vision first stage, the spectral knowledge-based, operational, near real-time Satellite Image Automatic Mapper™ (SIAM™) is selected from existing literature. To the best of these authors’ knowledge, this is the first time a symbolic syntactic inference system, like SIAM™, is made available to the RS community for operational use in a RS-IUS pre-attentive vision first stage, to accomplish multi-scale image segmentation and multi-granularity image pre-classification simultaneously, automatically and in near real-time. Full article