Inference Patterns Research Articles

An Infrastructural Brussels Effect: The translation of EU Law into the UK's digital borders

This article gives an account of the legal standards and safeguards that guide and constrain the current design of the UK's digital borders. Based on an empirical engagement with the development of Cerberus – an advanced risk-based analytics platform aimed at the detection of previously ‘unknown’ threats – the article presents a dual argument. On the one hand, it provides an account of the remaining salience and extraterritorial reach of EU law in setting standards for the collection, retention, processing and sharing of Passenger Name Records (PNR) data in the UK. This PNR data is a constitutive component of the digital border. Through the EU-UK Trade and Cooperation Agreement (TCA), the UK is now bound to comply with the rather stringent legal safeguards developed by the CJEU (in Opinion 1/15) on the retention and automated processing of PNR data. Our analysis shows the different channels through which EU law obtains this extraterritorial reach, how compliance can be monitored and enforced, and, crucially, how it has influenced and constrained the technical design of the UK's digital borders – a salient and unexplored phenomenon that we describe as an Infrastructural Brussels Effect. Yet, on the other hand, the article empirically shows that this is not merely a process of norm diffusion and extraterritoriality. Once legal standards become infrastructurally embedded in Cerberus, we witness normative translations and sociotechnical shifts with important legal and political consequences. Legal standards on ‘reasonable suspicion’ and the ‘objective evidence’ of ‘risk’, we argue, are given specific meaning through a logic of relational inference and algorithmic pattern detection (leading to forms of ‘concern by association’). By studying the entanglements between legal norms and material infrastructures – an approach we describe as infra-legalities – these normative effects become visible and contestable, providing a productive site for the sociolegal study of law and algorithmic governance.

Scales and inferences

Abstract Scalar inferences (SIs) are upper-bounding inferences associated with the use of semantically lower-bounded scalar expressions. One of the current debates regarding these inferences concerns their inferential pattern, specifically whether SIs are uniform or diverse across scales. This study follows the work on scalar diversity yet introduces two changes: First, we reexamine, from a different perspective, two structural properties of scales identified as accounting for SI diversity (boundedness and distance). Second, we analyze our data using both traditional regression analysis and complementary cluster analysis. The regression analysis demonstrates that our reexamination of the structural properties provides a more effective model, which also emphasizes the relationship between boundedness and distance. Specifically, we propose that boundedness fixes distance. The cluster analysis demonstrates two scale types: given-scales, which have an entrenched scalar construal, trigger SIs robustly; and volatile-scales, which have a fluctuant scalar construal, trigger SIs inconsistently. Building on these two scale types, we propose a necessary distinction between the conceptualization of a scale, which is diverse across different scales, and the actual derivation of the SI, which is uniform for all scales, once a scale has been construed. This distinction, we propose, explains how diversity can coexist alongside uniformity.

A fine-grained grading network for natural products based on dynamic association inference

Fine-grained grading of natural products, based on internal defects, plays an important role in food processing. Although X-ray imaging is critical for revealing internal defects due to its capacity to penetrate surfaces, the inherent grayscale and lower contrast make it challenging to extract fine-grained features. To address this, we propose the Fine-Grained Grading Network (FGGN) for classifying the quality grades of multiple natural products from X-ray images. Leveraging a ResNet18 backbone, FGGN enhances feature extraction by employing a hybrid attention module that emphasizes crucial local details by analyzing joint dependencies within tensor dimensions. Moreover, by combining cross-dimensional context encoding and dynamic graph convolution, a dynamic correlation inference module is proposed to construct dynamic graphs in a gated inference pattern, thereby integrating fine-grained information across abstraction levels. Finally, the cross-entropy loss is combined with graph constraint loss to train the network end-to-end. Extensive evaluation of three natural product datasets shows that the Fine-Grained Grading Network achieves state-of-the-art recognition performance, which demonstrates its grading applicability to multiple natural products.

Development of a chest X-ray machine learning convolutional neural network model on a budget and using artificial intelligence explainability techniques to analyze patterns of machine learning inference.

Machine learning (ML) will have a large impact on medicine and accessibility is important. This study's model was used to explore various concepts including how varying features of a model impacted behavior. This study built an ML model that classified chest X-rays as normal or abnormal by using ResNet50 as a base with transfer learning. A contrast enhancement mechanism was implemented to improve performance. After training with a dataset of publicly available chest radiographs, performance metrics were determined with a test set. The ResNet50 base was substituted with deeper architectures (ResNet101/152) and visualization methods used to help determine patterns of inference. Performance metrics were an accuracy of 79%, recall 69%, precision 96%, and area under the curve of 0.9023. Accuracy improved to 82% and recall to 74% with contrast enhancement. When visualization methods were applied and the ratio of pixels used for inference measured, deeper architectures resulted in the model using larger portions of the image for inference as compared to ResNet50. The model performed on par with many existing models despite consumer-grade hardware and smaller datasets. Individual models vary thus a single model's explainability may not be generalizable. Therefore, this study varied architecture and studied patterns of inference. With deeper ResNet architectures, the machine used larger portions of the image to make decisions. An example using a custom model showed that AI (Artificial Intelligence) can be accessible on consumer-grade hardware, and it also demonstrated an example of studying themes of ML explainability by varying ResNet architectures.

Pictorial syntax

It is commonly assumed that images, whether in the world or in the head, do not have a privileged analysis into constituent parts. They are thought to lack the sort of syntactic structure necessary for representing complex contents and entering into sophisticated patterns of inference. I reject this assumption. “Image grammars” are models in computer vision that articulate systematic principles governing the form and content of images. These models are empirically credible and can be construed as literal grammars for images. Images can have rich syntactic structure, though of a markedly different form than sentences in language.

Logic as a Science of Patterns?

I propose that logic may be seen as a science of patterns—however, not in the sense in which mathematics is a science of patterns, but rather in the sense in which physics is. The proposal is that logic identifies, explores, and fixes the inferential patterns which de facto govern our argumentative practices. It can be seen, I argue, as picking up the patterns and working from them toward the state of reflective equilibrium, where the laws it aims at are explicitly articulated. Due to the normativity of logic, this process is not quite the same as that in which a natural science works its way from data to its laws; however, there is no reason why it should be seen as proceeding via a mysterious "a priori analysis.” Instead, it should proceed in the intersubjective space in which the sciences are practiced.

Median-Joining Networks and Bayesian Phylogenies Often Do Not Tell the Same Story

Inferring phylogenies among intraspecific individuals often yields unresolved relationships (i.e., polytomies). Consequently, methods that compute distance-based abstract networks, like Median-Joining Networks (MJNs), are thought to be more appropriate tools for reconstructing such relationships than traditional trees. Median-Joining Networks visualize all routes of relationships in the form of cycles, if needed, when traditional approaches cannot resolve them. However, the MJN method is a distance-based phenetic approach that does not involve character transformations and makes no reference to ancestor–descendant relationships. Although philosophical and theoretical arguments challenging the implication that MJNs reflect phylogenetic signal in the traditional sense have been presented elsewhere, an empirical comparison with a character-based approach is needed given the increasing popularity of MJN analysis in evolutionary biology. Here, we use the conservative Approximately Unbiased (AU) test to compare 85 cases of branching patterns of cycle-free MJNs and Bayesian Inference (BI) phylogenies using datasets from 55 empirical studies. By rooting the MJN analyses to provide directionality, we report substantial disagreement between computed MJNs and posterior distributions on BI phylogenies. The branching patterns in MJNs and BI phylogenies show significantly different relationships in 37.6% of cases. Among the relationships that do not significantly differ, 96.2% show alternative sets of relationships. Our results indicate that the two methods provide different measures of relatedness in a phylogenetic sense. Finally, our analyses also support previous observations of the statistical hypothesis testing by reconfirming the over-conservativeness of the Shimodaira-Hasegawa test versus the AU test.

Modified Numerals and Split Disjunction: The First-Order Case

We present a number of puzzles arising for the interpretation of modified numerals. Following Büring and others we assume that the main difference between comparative and superlative modifiers is that only the latter convey disjunctive meanings. We further argue that the inference patterns triggered by disjunction and superlative modifiers are hard to capture in existing semantic and pragmatic analyses of these phenomena (neo-Gricean or grammatical alike), and we propose a novel account of these inferences in the framework of bilateral state-based modal logic defining a first order extension of Aloni (Semant Pragmat 15:5-EA, 2022. https://doi.org/10.3765/sp.15.5)’s BSML.

Differentiable learning of rules with constants in knowledge graph

Knowledge reasoning helps overcome the incompleteness of knowledge graphs (KGs) and has significantly contributed to the development of large KGs. Rule mining, one of the key tasks of knowledge reasoning, studies the problem of learning interpretable inference patterns over KGs. Existing rule mining methods mainly focus on learning rules that consist of different relations and variables, restricting the form of rules to be closed path. While rules could be diverse and in order to enrich the forms of rules, we argue that constants should also be considered in the rule mining process. We propose an Elegant Differentiable rUle learning with Constant mEthod (EduCe).11Source code of EduCeis available at https://github.com/yep96/Educe. We propose a constant operator and dynamic weight mechanism, which choose the constants that should be added and decrease the number of parameters, respectively. The model could mine diverse and accurate rules in an efficient way with these modules. The experimental results on several knowledge graph completion benchmarks show that EduCe achieves state-of-the-art link prediction results among differentiable rule mining methods and successfully learns diverse and high-quality rules.

John Eliot's Logick Primer: A Bilingual English-Massachusett Logic Textbook

In 1672 John Eliot, English Puritan educator and missionary to New England, published The Logick Primer: Some Logical Notions to initiate the INDIANS in the knowledge of the Rule of Reason; and to know how to make use thereof (Eliot 1672) The Logick Primer: Some Logical Notions to Initiate the INDIANS in the Knowledge of the Rule of Reason; and to Know How to Make Use Thereof , Cambridge, MA: Marmaduke Johnson]. This roughly 80 page pamphlet introduces syllogistic vocabulary and reasoning so that syllogisms can be created from Biblical texts. The use of logic for proselytizing purposes is not distinctive: What is distinctive about Eliot's book is that it is bilingual, written in both English and Massachusett (Wôpanâak), an Algonquian language spoken in eastern coastal and southeastern Massachusetts. It is one of the earliest bilingual logic textbooks and it is the first, and perhaps only, textbook in an indigenous American language. In this paper, we (1) introduce John Eliot and the linguistic context he was working in; (2) introduce the contents of the Logick Primer – vocabulary, inference patterns, and applications; (3) discuss notions of ‘Puritan’ logic that inform this primer; and (4) address the importance of his work in documenting and expanding the Massachusett language and the problems that accompany his colonial approach to this work.

LOGICAL SCALES IMAM AL-GHAZALI (505 AH) AND SOME OF ITS APPLICATIONS TO MATTERS OF BELIEF


 
 
 
 Imam Al-Ghazali, may God have mercy on him, tried to make use of the logical rules to serve the science of belief, and it is no secret that logical measurement is one of the most accurate sections of this science. They are: Equilibrium Balance, which is: Returning the first form of gestational measurement to the Greater Equilibrium Balance, it is divided into three sections (the largest, middle, and smallest balance) and the second balance is the balance of concomitance and this pattern of inference is organized within the framework of the connected conditional measurement that consists of a conditional introduction connected to Zou This is a measure of inference in the framework of a separate conditional measurement, which is unlike the pattern of convergence, and consists of a separate conditional introduction and a local introduction and a local result. 
 
 
 
 

For when you just can't talk to 'normal' people . . . Exploring the use of informal support structures by supernumerary university paramedic students: findings from a phenomenological study.

In an effort to shield them from distress, pre-hospital ambulance staff may avoid discussing traumatic workplace experiences with friends and family. As a source of informal support, however, workplace camaraderie is considered important for managing occupational stress. For supernumerary university paramedic students there is limited research concerning how such experiences are managed, and whether they may benefit from similar informal support. This is a concerning deficit when contextualised by reports of higher stress levels among students undertaking work-based learning, and among paramedics / paramedic students in general. These original findings allude to how university paramedic students who are supernumerary within the pre-hospital workplace utilise informal support mechanisms. A qualitative, interpretive approach was adopted. University paramedic students were recruited via purposive sampling. Audio-recorded face-to-face semi-structured interviews were performed and transcribed verbatim. Analysis involved initial descriptive coding and then inferential pattern coding. The identification of themes and discussion topics was facilitated by a review of the literature. Twelve participants were recruited aged from 19 to 27 years, and 58% (n = 7) were female. While most participants cited that they were able to enjoy the informal stress-relieving camaraderie of ambulance staff, there were perceptions that supernumerary status may leave them potentially isolated within the workplace. Participants may also compartmentalise their experiences away from friends and family in a manner similar to that found among ambulance staff. Informal student peer support networks were praised as a source of information and for emotional support. Self-organised online chat groups were ubiquitous as a means of keeping in touch with student peers. While undertaking pre-hospital practice placements, supernumerary university paramedic students may not have complete access to the informal support of ambulance staff, and they may feel unable to discuss stressful feelings with friends or loved ones. However, within this study, self-moderated online chat groups were used almost universally as a readily accessible means of peer support. Paramedic educators ideally need an awareness of how such groups are used to ensure that they represent a supportive and inclusive space for students. Further research into how university paramedic students utilise online chat groups for peer support may further reveal a potentially valuable informal support structure.

Dehumanization as a Response to Uncivil and Immoral Behaviors.

Theoretical approaches to dehumanization consider civility to be an attribute of human uniqueness (HU). However, studies that explore the links between civility and humanness are scarce. More precisely, the present research tests whether there is a consistent relationship between civility and HU. Method and results: The first study (N = 192; Mage = 19.91; SD = 2.70; 69% women) shows that individuals infer more HU traits in the agents of civil behaviors compared to agents of other positive behaviors that are not related to civility. The second study (N = 328; Mage = 19.69; SD = 3.65; 77% women) reveals that uncivil and immoral behaviors displayed a similar pattern of inference of HU traits; however, moral behaviors were more associated with human nature than civil behaviors. Conclusions: Overall, results confirmed that civil behaviors facilitate the inference of humanness, specifically of HU traits, and that civil and moral behaviors are not equivalent in terms of the human inferences to which they lead.

Genomic signatures and evolutionary history of the endangered blue-crowned laughingthrush and other Garrulax species

BackgroundThe blue-crowned laughingthrush (Garrulax courtoisi) is a critically endangered songbird endemic to Wuyuan, China, with population of ~323 individuals. It has attracted widespread attention, but the lack of a published genome has limited research and species protection.ResultsWe report two laughingthrush genome assemblies and reveal the taxonomic status of laughingthrush species among 25 common avian species according to the comparative genomic analysis. The blue-crowned laughingthrush, black-throated laughingthrush, masked laughingthrush, white-browed laughingthrush, and rusty laughingthrush showed a close genetic relationship, and they diverged from a common ancestor between ~2.81 and 12.31 million years ago estimated by the population structure and divergence analysis using 66 whole-genome sequencing birds from eight laughingthrush species and one out group (Cyanopica cyanus). Population inference revealed that the laughingthrush species experienced a rapid population decline during the last ice age and a serious bottleneck caused by a cold wave during the Chinese Song Dynasty (960–1279 AD). The blue-crowned laughingthrush is still in a bottleneck, which may be the result of a cold wave together with human exploitation. Interestingly, the existing blue-crowned laughingthrush exhibits extremely rich genetic diversity compared to other laughingthrushes. These genetic characteristics and demographic inference patterns suggest a genetic heritage of population abundance in the blue-crowned laughingthrush. The results also suggest that fewer deleterious mutations in the blue-crowned laughingthrush genomes have allowed them to thrive even with a small population size. We believe that cooperative breeding behavior and a long reproduction period may enable the blue-crowned laughingthrush to maintain genetic diversity and avoid inbreeding depression. We identified 43 short tandem repeats that can be used as markers to identify the sex of the blue-crowned laughingthrush and aid in its genetic conservation.ConclusionsThis study supplies the missing reference genome of laughingthrush, provides insight into the genetic variability, evolutionary potential, and molecular ecology of laughingthrush and provides a genomic resource for future research and conservation.

Epistemic Paternalism via Conceptual Engineering

AbstractThis essay focuses on conceptual engineers who aim to improve other people's patterns of inference and attention by shaping their concepts. Such conceptual engineers sometimes engage in a form of epistemic paternalism that I call paternalistic cognitive engineering: instead of explicitly persuading, informing and educating others, the engineers non-consultatively rely on assumptions about the target agents’ cognitive systems to improve their belief forming. The target agents could reasonably regard such benevolent exercises of control as violating their sovereignty over their own belief formation. This is a pro tanto reason against such engineering. In addition to the relevant projects of conceptual engineering, paternalistic cognitive engineering plausibly includes certain kinds of nudging and evidence suppression. I distinguish the sovereignty-based concern from other ethical worries about conceptual engineering and discuss how one might justify the relevant conceptual engineering projects despite the sovereignty-based reason against them.

A note on the normality assumption for Bayesian models of constraint in behavioral individual differences

To investigate the structure of individual differences in performance on behavioral tasks, Haaf and Rouder (2017) developed a class of hierarchical Bayesian mixed models with varying levels of constraint on the individual effects. The models are then compared via Bayes factors, telling us which model best predicts the observed data. One common criticism of their method is that the observed data are assumed to be drawn from a normal distribution. However, for most cognitive tasks, the primary measure of performance is a response time, the distribution of which is well known not to be normal. In this paper, I investigate the assumption of normality for two datasets in numerical cognition. Specifically, I show that using a shifted lognormal model for the response times does not change the overall pattern of inference. Further, since the model-estimated effects are now on a logarithmic scale, the interpretation of the modeling becomes more difficult, particularly because the estimated effect is now multiplicative rather than additive. As a result, I recommend that even though response times are not normally distributed in general, the simplification afforded by the Haaf and Rouder's approach provides a pragmatic approach to modeling individual differences in behavioral tasks.

Geometry Interaction Knowledge Graph Embeddings

Knowledge graph (KG) embeddings have shown great power in learning representations of entities and relations for link prediction tasks. Previous work usually embeds KGs into a single geometric space such as Euclidean space (zero curved), hyperbolic space (negatively curved) or hyperspherical space (positively curved) to maintain their specific geometric structures (e.g., chain, hierarchy and ring structures). However, the topological structure of KGs appears to be complicated, since it may contain multiple types of geometric structures simultaneously. Therefore, embedding KGs in a single space, no matter the Euclidean space, hyperbolic space or hyperspheric space, cannot capture the complex structures of KGs accurately. To overcome this challenge, we propose Geometry Interaction knowledge graph Embeddings (GIE), which learns spatial structures interactively between the Euclidean, hyperbolic and hyperspherical spaces. Theoretically, our proposed GIE can capture a richer set of relational information, model key inference patterns, and enable expressive semantic matching across entities. Experimental results on three well-established knowledge graph completion benchmarks show that our GIE achieves the state-of-the-art performance with fewer parameters.

Hybrid Autoregressive Inference for Scalable Multi-Hop Explanation Regeneration

Regenerating natural language explanations in the scientific domain has been proposed as a benchmark to evaluate complex multi-hop and explainable inference. In this context, large language models can achieve state-of-the-art performance when employed as cross-encoder architectures and fine-tuned on human-annotated explanations. However, while much attention has been devoted to the quality of the explanations, the problem of performing inference efficiently is largely under studied. Cross-encoders, in fact, are intrinsically not scalable, possessing limited applicability to real-world scenarios that require inference on massive facts banks. To enable complex multi-hop reasoning at scale, this paper focuses on bi-encoder architectures, investigating the problem of scientific explanation regeneration at the intersection of dense and sparse models. Specifically, we present SCAR (for Scalable Autoregressive Inference), a hybrid framework that iteratively combines a Transformer-based bi-encoder with a sparse model of explanatory power, designed to leverage explicit inference patterns in the explanations. Our experiments demonstrate that the hybrid framework significantly outperforms previous sparse models, achieving performance comparable with that of state-of-the-art cross-encoders while being approx 50 times faster and scalable to corpora of millions of facts. Further analyses on semantic drift and multi-hop question answering reveal that the proposed hybridisation boosts the quality of the most challenging explanations, contributing to improved performance on downstream inference tasks.

Link Prediction in Knowledge Graphs: A Hierarchy-Constrained Approach

Link prediction over a knowledge graph aims to predict the missing head entities <inline-formula><tex-math notation="LaTeX">$h$</tex-math></inline-formula> or tail entities <inline-formula><tex-math notation="LaTeX">$t$</tex-math></inline-formula> and missing relations <inline-formula><tex-math notation="LaTeX">$r$</tex-math></inline-formula> for a triple <inline-formula><tex-math notation="LaTeX">$(h,r,t)$</tex-math></inline-formula> . Recent years have witnessed great advance of knowledge graph embedding based link prediction methods, which represent entities and relations as elements of a continuous vector space. Most methods learn the embedding vectors by optimizing a margin-based loss function, where the margin is used to separate negative and positive triples in the loss function. The loss function utilizes the general structures of knowledge graphs, e.g., the vector of <inline-formula><tex-math notation="LaTeX">$r$</tex-math></inline-formula> is the translation of the vector of <inline-formula><tex-math notation="LaTeX">$h$</tex-math></inline-formula> and <inline-formula><tex-math notation="LaTeX">$t$</tex-math></inline-formula> , and the vector of <inline-formula><tex-math notation="LaTeX">$t$</tex-math></inline-formula> should be the nearest neighbor of the vector of <inline-formula><tex-math notation="LaTeX">$h+r$</tex-math></inline-formula> . However, there are many particular structures, and can be employed to promote the performance of link prediction. One typical structure in knowledge graphs is hierarchical structure, which existing methods have much unexplored. We argue that the hierarchical structures also contain rich inference patterns, and can further enhance the link prediction performance. In this paper, we propose a hierarchy-constrained link prediction method, called hTransM, on the basis of the translation-based knowledge graph embedding methods. It can adaptively determine the optimal margin by detecting the single-step and multi-step hierarchical structures. Moreover, we prove the effectiveness of hTransM theoretically, and experiments over three benchmark datasets and two sub-tasks of link prediction demonstrate the superiority of hTransM.

An Active Inference Account of Touch and Verbal Communication in Therapy.

This paper offers theoretical explanations for why “guided touch” or manual touch with verbal communication can be an effective way of treating the body (e.g., chronic pain) and the mind (e.g., emotional disorders). The active inference theory suggests that chronic pain and emotional disorders can be attributed to distorted and exaggerated patterns of interoceptive and proprioceptive inference. We propose that the nature of active inference is abductive. As such, to rectify aberrant active inference processes, we should change the “Rule” of abduction, or the “prior beliefs” entailed by a patient’s generative model. This means pre-existing generative models should be replaced with new models. To facilitate such replacement—or updating—the present treatment proposes that we should weaken prior beliefs, especially the one at the top level of hierarchical generative models, thereby altering the sense of agency, and redeploying attention. Then, a new prior belief can be installed through inner communication along with manual touch. The present paper proposes several hypotheses for possible experimental studies. If touch with verbal guidance is proven to be effective, this would demonstrate the relevance of active inference and the implicit prediction model at a behavioral level. Furthermore, it would open new possibilities of employing inner communication interventions, including self-talk training, for a wide range of psychological and physical therapies.