Context Of Inference Research Articles

Towards responsible AI: an implementable blueprint for integrating explainability and social-cognitive frameworks in AI systems

Automation increasingly shapes modern society, requiring artificial intelligence (AI) systems to not only perform complex tasks but also provide clear, actionable explanations of their decisions, especially in high-stakes domains. However, most contemporary AI systems struggle to explain their runtime operations in specific instances, limiting their applicability in contexts demanding stringent outcome justification. Existing approaches have attempted to address this challenge but often fall short in terms of contextual relevance, human cognitive alignment, or scalability. This paper introduces System-of-Systems Machine Learning (SoS-ML) as a novel framework to advance explainable artificial intelligence (XAI) by addressing the limitations of current methods. Drawing from insights in philosophy, cognitive science, and social sciences, SoS-ML seeks to integrate human-like reasoning processes into AI, framing explanations as contextual inferences and justifications. The research demonstrates how SoS-ML addresses key challenges in XAI, such as enhancing explanation accuracy and aligning AI reasoning with human cognition. By leveraging a multi-agent, modular design, SoS-ML encourages collaboration among machine learning models, leading to more transparent, context-aware systems. The framework’s ability to generalize across domains is demonstrated through experiments on the Pima Indian Diabetes dataset and pie chart image-to-text interpretation, showcasing its transformative potential in improving both model accuracy and explainability. The findings emphasize SoS-ML’s role in advancing responsible AI, particularly in high-stakes environments where interpretability and social accountability are paramount.

A generative adversarial network for inferring age attributes in subway based on automatic fare collection and point of interest data

Understanding the socioeconomic characteristics of passengers is crucial for intelligent services for subway systems. Nevertheless, obtaining such information poses challenges, especially sensitive attributes such as age, due to heightened considerations for data privacy. To address this challenge, we explore the correlation between travel features and age from both macro and micro perspectives and propose an improved Generative Adversarial Network (GAN) for age inference in subway contexts. In the macro level, we extract statistical features from Automatic Fare Collection data and novel features from Points of Interest data. The novel features allow us to examine the relationship between built environment and age. In the micro level, we introduce a comprehensive travel chain, considering the dynamic evolution of passenger trips during multiple days. To address the sample imbalance across age groups, we propose an improved GAN-based framework to infer the age attribute. Our framework incorporates Inner Product to capture interdependency among macro features and employs a Transformer-based module to extract sequential and spatiotemporal features within micro-level travel chains. Finally, the framework is validated using a case study of Guangzhou subway. The results indicate the superiority of our model over baseline approaches like Random Forest, Bayesian Networks, and Variational Autoencoders, showcasing accuracy improvements of 17.75%, 20.29%, and 16.10%, respectively. The use of Inner Product and Transformer-based modules enhances the predictive accuracy of the model. The utilization of travel chains proves to be effective in capturing travel features and improves the model’s accuracy in age inference. Furthermore, we delve into potential applications.

Consistent causal inference for high-dimensional time series

Consistent causal inference for high-dimensional time series

Contextualised retrieval practice: the effect of context type and working memory capacity on word retention

ABSTRACT The present study on intentional retrieval practice compared the benefits of presenting words in either informative or uninformative sentence contexts. Participants first studied a list of English words with their translations. Then, they were all exposed to half of the words with informative sentences containing meaning clues in the Context Inference (CI) condition and half with uninformative sentences devoid of such clues in the Memory Retrieval (MR) condition as part of retrieval-based practising. Participants were required to type the L1 translation for each word presented using a mobile application. Data were collected by both form-recall and meaning-recall tests immediately afterwards and then a week later. In addition, this study focused on the relationship between working memory capacity (WMC) and word retention in these two conditions to explore the suggestion that individuals may benefit differently from retrieval practices. Although the results showed that both conditions contributed to word retention, the MR condition was significantly more effective than the CI condition for the participants’ long-term retention. Further, the results revealed an overall positive effect of WMC on word retention in both conditions, with high-WMC individuals achieving higher retention scores than low-WMC individuals. However, this effect was not modulated by the type of context condition.

Dynamics-Adaptive Continual Reinforcement Learning via Progressive Contextualization.

A key challenge of continual reinforcement learning (CRL) in dynamic environments is to promptly adapt the reinforcement learning (RL) agent's behavior as the environment changes over its lifetime while minimizing the catastrophic forgetting of the learned information. To address this challenge, in this article, we propose DaCoRL, that is, dynamics-adaptive continual RL. DaCoRL learns a context-conditioned policy using progressive contextualization, which incrementally clusters a stream of stationary tasks in the dynamic environment into a series of contexts and opts for an expandable multihead neural network to approximate the policy. Specifically, we define a set of tasks with similar dynamics as an environmental context and formalize context inference as a procedure of online Bayesian infinite Gaussian mixture clustering on environment features, resorting to online Bayesian inference to infer the posterior distribution over contexts. Under the assumption of a Chinese restaurant process (CRP) prior, this technique can accurately classify the current task as a previously seen context or instantiate a new context as needed without relying on any external indicator to signal environmental changes in advance. Furthermore, we employ an expandable multihead neural network whose output layer is synchronously expanded with the newly instantiated context and a knowledge distillation regularization term for retaining the performance on learned tasks. As a general framework that can be coupled with various deep RL algorithms, DaCoRL features consistent superiority over existing methods in terms of stability, overall performance, and generalization ability, as verified by extensive experiments on several robot navigation and MuJoCo locomotion tasks.

Rapid context inference in a thalamocortical model using recurrent neural networks

Cognitive flexibility is a fundamental ability that enables humans and animals to exhibit appropriate behaviors in various contexts. The thalamocortical interactions between the prefrontal cortex (PFC) and the mediodorsal thalamus (MD) have been identified as crucial for inferring temporal context, a critical component of cognitive flexibility. However, the neural mechanism responsible for context inference remains unknown. To address this issue, we propose a PFC-MD neural circuit model that utilizes a Hebbian plasticity rule to support rapid, online context inference. Specifically, the model MD thalamus can infer temporal contexts from prefrontal inputs within a few trials. This is achieved through the use of PFC-to-MD synaptic plasticity with pre-synaptic traces and adaptive thresholding, along with winner-take-all normalization in the MD. Furthermore, our model thalamus gates context-irrelevant neurons in the PFC, thus facilitating continual learning. We evaluate our model performance by having it sequentially learn various cognitive tasks. Incorporating an MD-like component alleviates catastrophic forgetting of previously learned contexts and demonstrates the transfer of knowledge to future contexts. Our work provides insight into how biological properties of thalamocortical circuits can be leveraged to achieve rapid context inference and continual learning.

Comparison of micro- and macro- structural narrative features between Turkish-Kurdish bilinguals and Turkish monolinguals

Abstract Bilingualism may lead to differences in narrative micro- and macrostructural features. Creating a normative data set on the narrative characteristics of healthy Turkish-Kurdish bilinguals may allow for a more effective assessment of language disorders that may occur in bilinguals with acquired brain injury. This study aims to determine how narrative micro- and macrostructural features differ in Turkish-Kurdish bilinguals in single picture and story narratives. A total of 23 bilingual (Turkish-Kurdish) and 23 monolingual (Turkish) healthy individuals were included in the study. Narrative samples were obtained using single pictures and a story. Micro- and macrostructural features were compared between bilinguals and monolinguals matched for age, education, and gender. Mean length utterance-morpheme and clausal density variables, considered as indicators of grammatical complexity, were higher in monolinguals in both single picture and story narratives. Morphemic errors were higher for bilinguals. Effort behaviors, which can interrupt verbal efficiency, were higher in bilinguals in single picture narratives. While there was no difference between the groups in verbal fluency (phonemic and categorical), the lexical diversty in storytelling was higher in monolinguals. In terms of macrostructural features, cognitive inferences in the story context and emotional discourse in the single picture context were higher in monolinguals, while enrichment expressions in the single picture narrative were higher in bilinguals. In both groups, macrostructural features such as cognitive inferences and expressions of uncertainty in discourse were negatively correlated with grammatical complexity variables. Turkish-Kurdish bilinguals’ single picture and story narratives in Turkish may show semantic, morphosyntactic and pragmatic differences compared to Turkish-speaking monolinguals.

Different mechanisms of contextual inference govern associatively learned and sensory-evoked postural responses

Human standing balance relies on the continuous monitoring and integration of sensory signals to infer our body's motion and orientation within the environment. However, when sensory information is no longer contextually relevant to balancing the body (e.g., when sensory and motor signals are incongruent), sensory-evoked balance responses are rapidly suppressed, much earlier than any conscious perception of changes in balance control. Here, we used a robotic balance simulator to assess whether associatively learned postural responses are similarly modulated by sensorimotor incongruence and contextual relevance to postural control. Twenty-nine participants in three groups were classically conditioned to generate postural responses to whole-body perturbations when presented with an initially neutral sound cue. During catch and extinction trials, participants received only the auditory stimulus but in different sensorimotor states corresponding to their group: 1) during normal active balance, 2) while immobilized, and 3) throughout periods where the computer subtly removed active control over balance. In the balancing and immobilized states, conditioned responses were either evoked or suppressed, respectively, according to the (in)ability to control movement. Following the immobilized state, conditioned responses were renewed when balance was restored, indicating that conditioning was retained but only expressed when contextually relevant. In contrast, conditioned responses persisted in the computer-controlled state even though there was no causal relationship between motor and sensory signals. These findings suggest that mechanisms responsible for sensory-evoked and conditioned postural responses do not share a single, central contextual inference and assessment of their relevance to postural control, and may instead operate in parallel.

Decision uncertainty as a context for motor memory

The current view of perceptual decision-making suggests that once a decision is made, only a single motor programme associated with the decision is carried out, irrespective of the uncertainty involved in decision making. In contrast, we show that multiple motor programmes can be acquired on the basis of the preceding uncertainty of the decision, indicating that decision uncertainty functions as a contextual cue for motor memory. The actions learned after making certain (uncertain) decisions are only partially transferred to uncertain (certain) decisions. Participants were able to form distinct motor memories for the same movement on the basis of the preceding decision uncertainty. Crucially, this contextual effect generalizes to novel stimuli with matched uncertainty levels, demonstrating that decision uncertainty is itself a contextual cue. These findings broaden the understanding of contextual inference in motor memory, emphasizing that it extends beyond direct motor control cues to encompass the decision-making process.

Contingency inferences from base rates: A parsimonious strategy?

The pseudocontingency framework provides a parsimonious strategy for inferring the contingency between two variables by assessing the base rates. Frequently occurring levels are associated, as are rarely occurring levels. However, this strategy can lead to different contingency inferences in different contexts, depending on how the base rates vary across contexts. Here, we examine how base-rate consistency influences base-rate learning and reliance by contrasting consistent with inconsistent base rates. We hypothesized that base-rate learning is facilitated, and that people rely more on base rates if base rates are consistent. In Experiment 1, the base rates across four contexts implied the same (consistent) or different (inconsistent) contingencies. Base rates were learned equally accurately, and participants inferred contingencies that followed the base rates but deviated from the genuine contingencies within contexts, regardless of consistency. In Experiment 2, we additionally manipulated whether the context was a plausible moderator of the contingency. While we replicated the first experiment's results when the context was a plausible moderator, base-rate inferences were stronger for consistent base rates when the context was an implausible moderator. Possibly, when a moderation-by-context was implausible, participants also relied on the base-rate correlation across contexts, which implied the same contingency when base rates were consistent but was zero when the base rates were inconsistent. Thus, our findings suggest that contingency inferences from base rates involve top-down processes in which people decide how to use base-rate information.

ConvoSense: Overcoming Monotonous Commonsense Inferences for Conversational AI

Abstract Mastering commonsense understanding and reasoning is a pivotal skill essential for conducting engaging conversations. While there have been several attempts to create datasets that facilitate commonsense inferences in dialogue contexts, existing datasets tend to lack in-depth details, restate information already present in the conversation, and often fail to capture the multifaceted nature of commonsense reasoning. In response to these limitations, we compile a new synthetic dataset for commonsense reasoning in dialogue contexts using GPT, ℂonvoSense, that boasts greater contextual novelty, offers a higher volume of inferences per example, and substantially enriches the detail conveyed by the inferences. Our dataset contains over 500,000 inferences across 12,000 dialogues with 10 popular inference types, which empowers the training of generative commonsense models for dialogue that are superior in producing plausible inferences with high novelty when compared to models trained on the previous datasets. To the best of our knowledge, ℂonvoSense is the first of its kind to provide such a multitude of novel inferences at such a large scale.

Frustratives in St’át’imcets vs. Kimaragang: parameters of variation

Markers of the FRUSTRATIVE (non-realization of an expected outcome) often have a variety of uses. The Kimaragang frustrative dara also marks various related functions such as “discontinuous past”, “action narrowly averted”, frustrated intention or desire, and counterfactual conditionals. Davis & Matthewson (2022) describe a similar but not identical range of uses for the frustrative marker séna7 in St’át’imcets, and propose that the core meaning of this marker involves a type of epistemic modality. They further propose that their analysis can be extended to Kimaragang dara, suggesting that observed differences of usage may be due to differences in the tense-aspect systems of the two languages.This paper argues that the Kimaragang frustrative is different from séna7 in two respects: (1) the nature of the “frustrated expectation” inference, and (2) type of modality (epistemic vs. circumstantial). Frustrated expectation is lexically entailed in St’át’imcets, but Kimaragang dara entails only that the expected outcome cannot be asserted to be true at the time of speaking, with frustrated expectation arising as a pragmatic inference in contexts where it is reasonable to expect the speaker to know the actual outcome. Davis & Matthewson show that séna7 is purely epistemic in nature: it marks an unexpected correlation between two propositions, which need not belong to a single chain of events. Kimaragang dara is circumstantial in nature: it marks an initial situation whose expected outcome within the normal course of events is so far unrealized. It can be used to describe counterfactual situations, which are not epistemically accessible.

GCN-assisted attention-guided UNet for automated retinal OCT segmentation

GCN-assisted attention-guided UNet for automated retinal OCT segmentation

MetaKernel: Learning Variational Random Features With Limited Labels.

Few-shot learning deals with the fundamental and challenging problem of learning from a few annotated samples, while being able to generalize well on new tasks. The crux of few-shot learning is to extract prior knowledge from related tasks to enable fast adaptation to a new task with a limited amount of data. In this paper, we propose meta-learning kernels with random Fourier features for few-shot learning, we call MetaKernel. Specifically, we propose learning variational random features in a data-driven manner to obtain task-specific kernels by leveraging the shared knowledge provided by related tasks in a meta-learning setting. We treat the random feature basis as the latent variable, which is estimated by variational inference. The shared knowledge from related tasks is incorporated into a context inference of the posterior, which we achieve via a long-short term memory module. To establish more expressive kernels, we deploy conditional normalizing flows based on coupling layers to achieve a richer posterior distribution over random Fourier bases. The resultant kernels are more informative and discriminative, which further improves the few-shot learning. To evaluate our method, we conduct extensive experiments on both few-shot image classification and regression tasks. A thorough ablation study demonstrates that the effectiveness of each introduced component in our method. The benchmark results on fourteen datasets demonstrate MetaKernel consistently delivers at least comparable and often better performance than state-of-the-art alternatives.

CoBjeason: Reasoning Covered Object in Image by Multi-Agent Collaboration Based on Informed Knowledge Graph

Object detection is a widely studied problem in existing works. However, in this paper, we turn to a more challenging problem of “ Covered Object Reasoning ”, aimed at reasoning the category label of target object in the given image particularly when it has been totally covered (or invisible ). To resolve this problem, we propose CoBjeason to seize the opportunity when visual reasoning meets the knowledge graph, where “ empirical cognition ” on common visual contexts have been incorporated as knowledge graph to conduct reinforced multi-hop reasoning via two collaborative agents. Such two agents, for one thing, stand at the covered object (or unknown entity ) to observe the surrounding visual cues in the given image and gradually select entities and relations from the global gallery-level knowledge graph which contains entity-pairs frequently occurring across the entire image-collection, so as to infer the main structure of image-level knowledge graph forward expanded from the unknown entity . In turn, for another, based on the reasoned image-level knowledge graph, the semantic context among entities will be aggregated backward into unknown entity to select an appropriate entity from the global gallery-level knowledge graph as the reasoning result. Moreover, such two agents will collaborate with each other, securing that the above Forward & Backward Reasoning will step towards the same destination of the higher performance on covered object reasoning. To our best knowledge, this is the first work on Covered Object Reasoning with Knowledge Graphs and reinforced Multi-Agent collaboration. Particularly, our study on Covered Object Reasoning and the proposed model CoBjeason could offer novel insights into more basic Computer Vision (CV) tasks, such as Semantic Segmentation with better understanding on the current scene when some objects are blurred or covered, Visual Question Answering with enhancement on the inference in more complicated visual context when some objects are covered or invisible, and Image Caption Generation with the augmentation on the richness of visual context for images containing partially visible objects. The improvement on the above basic CV tasks can further refine more complicated ones involved with nuanced visual interpretation like Autonomous Driving, where the recognition and reasoning on partially visible or covered object are critical. According to the experimental results, our proposed CoBjeason can achieve the best overall ranking performance on covered object reasoning compared with other models, meanwhile enjoying the advantage of lower “ exploration cost ”, with the insensitivity against the long-tail covered objects and the acceptable time complexity.

A machine learning approach for real-time cortical state estimation

Objective. Cortical function is under constant modulation by internally-driven, latent variables that regulate excitability, collectively known as ‘cortical state’. Despite a vast literature in this area, the estimation of cortical state remains relatively ad hoc, and not amenable to real-time implementation. Here, we implement robust, data-driven, and fast algorithms that address several technical challenges for online cortical state estimation. Approach. We use unsupervised Gaussian mixture models to identify discrete, emergent clusters in spontaneous local field potential signals in cortex. We then extend our approach to a temporally-informed hidden semi-Markov model (HSMM) with Gaussian observations to better model and infer cortical state transitions. Finally, we implement our HSMM cortical state inference algorithms in a real-time system, evaluating their performance in emulation experiments. Main results. Unsupervised clustering approaches reveal emergent state-like structure in spontaneous electrophysiological data that recapitulate arousal-related cortical states as indexed by behavioral indicators. HSMMs enable cortical state inferences in a real-time context by modeling the temporal dynamics of cortical state switching. Using HSMMs provides robustness to state estimates arising from noisy, sequential electrophysiological data. Significance. To our knowledge, this work represents the first implementation of a real-time software tool for continuously decoding cortical states with high temporal resolution (40 ms). The software tools that we provide can facilitate our understanding of how cortical states dynamically modulate cortical function on a moment-by-moment basis and provide a basis for state-aware brain machine interfaces across health and disease.

Fuzzy information recognition and translation processing in English interpretation based on a generalized maximum likelihood ratio algorithm.

English interpretation plays a vital role as a critical link in cross-language communication. However, there are various types of ambiguous information in many interpreting scenarios, such as ambiguity, ambiguous vocabulary, and syntactic structures, which may lead to inaccuracies and fluency issues in translation. This article proposes a method based on the generalized maximum likelihood ratio algorithm (GLR) to identify and process fuzzy information in English interpretation to improve the quality and efficiency of performance. Firstly, we systematically analyzed the common types of fuzzy information in interpretation and delved into the basic principles and applications of the generalized maximum likelihood ratio algorithm. This algorithm is widely used in natural language processing to solve uncertainty problems and has robust modeling and inference capabilities, making it suitable for handling fuzzy information in interpretation. Then, we propose a fuzzy information recognition model based on the generalized maximum likelihood ratio algorithm. This model utilizes a large-scale interpretation corpus for training and identifies potential fuzzy information in the interpretation process through statistical analysis and pattern recognition. Once fuzzy information is detected, we adopt a series of effective translation processing strategies, including contextual inference and adaptation, to ensure the accuracy and naturalness of interpretation. Finally, we conducted a series of experiments to evaluate the performance of the proposed method. The experimental results show that the fuzzy information recognition and translation processing method based on the generalized maximum likelihood ratio algorithm performs well in different interpretation scenarios, significantly improving the quality and fluency of interpretation and reducing ambiguity caused by fuzzy information.

Dual-track spatio-temporal learning for urban flow prediction with adaptive normalization

Dual-track spatio-temporal learning for urban flow prediction with adaptive normalization

Meta-learning with deep flow kernel network for few shot cross-domain remaining useful life prediction

Meta-learning with deep flow kernel network for few shot cross-domain remaining useful life prediction

Tomb mural inpainting with adaptive convolutional constraints and global context inference

Tomb mural inpainting with adaptive convolutional constraints and global context inference