Jesse Hoey

Blake VanBerlo, Brian Li, Alexander Wong et al.

Self-supervised pretraining has been observed to improve performance in supervised learning tasks in medical imaging. This study investigates the utility of self-supervised pretraining prior to conducting supervised fine-tuning for the downstream task of lung sliding classification in M-mode lung ultrasound images. We propose a novel pairwise relationship that couples M-mode images constructed from the same B-mode image and investigate the utility of data augmentation procedure specific to M-mode lung ultrasound. The results indicate that self-supervised pretraining yields better performance than full supervision, most notably for feature extractors not initialized with ImageNet-pretrained weights. Moreover, we observe that including a vast volume of unlabelled data results in improved performance on external validation datasets, underscoring the value of self-supervision for improving generalizability in automatic ultrasound interpretation. To the authors' best knowledge, this study is the first to characterize the influence of self-supervised pretraining for M-mode ultrasound.

Blake VanBerlo, Brian Li, Jesse Hoey et al.

In this study, we investigated whether self-supervised pretraining could produce a neural network feature extractor applicable to multiple classification tasks in B-mode lung ultrasound analysis. When fine-tuning on three lung ultrasound tasks, pretrained models resulted in an improvement of the average across-task area under the receiver operating curve (AUC) by 0.032 and 0.061 on local and external test sets respectively. Compact nonlinear classifiers trained on features outputted by a single pretrained model did not improve performance across all tasks; however, they did reduce inference time by 49% compared to serial execution of separate fine-tuned models. When training using 1% of the available labels, pretrained models consistently outperformed fully supervised models, with a maximum observed test AUC increase of 0.396 for the task of view classification. Overall, the results indicate that self-supervised pretraining is useful for producing initial weights for lung ultrasound classifiers.

Blake VanBerlo, Jesse Hoey, Alexander Wong

Self-supervised pretraining has been observed to be effective at improving feature representations for transfer learning, leveraging large amounts of unlabelled data. This review summarizes recent research into its usage in X-ray, computed tomography, magnetic resonance, and ultrasound imaging, concentrating on studies that compare self-supervised pretraining to fully supervised learning for diagnostic tasks such as classification and segmentation. The most pertinent finding is that self-supervised pretraining generally improves downstream task performance compared to full supervision, most prominently when unlabelled examples greatly outnumber labelled examples. Based on the aggregate evidence, recommendations are provided for practitioners considering using self-supervised learning. Motivated by limitations identified in current research, directions and practices for future study are suggested, such as integrating clinical knowledge with theoretically justified self-supervised learning methods, evaluating on public datasets, growing the modest body of evidence for ultrasound, and characterizing the impact of self-supervised pretraining on generalization.

Jesse Hoey, Gabrielle Chan

Any decision, such as one about who to hire, involves two components. First, a rational component, i.e., they have a good education, they speak clearly. Second, an affective component, based on observables such as visual features of race and gender, and possibly biased by stereotypes. Here we propose a method for measuring the affective, socially biased, component, thus enabling its removal. That is, given a decision-making process, these affective measurements remove the affective bias in the decision, rendering it fair across a set of categories defined by the method itself. We thus propose that this may solve three key problems in intersectional fairness: (1) the definition of categories over which fairness is a consideration; (2) an infinite regress into smaller and smaller groups; and (3) ensuring a fair distribution based on basic human rights or other prior information. The primary idea in this paper is that fairness biases can be measured using affective coherence, and that this can be used to normalize outcome mappings. We aim for this conceptual work to expose a novel method for handling fairness problems that uses emotional coherence as an independent measure of bias that goes beyond statistical parity.

Misha Melnyk, Mitchell Dolny, Joshua D. Elkind et al.

Policy setting in the USA according to the ``Garbage Can'' model differentiates between ``problem'' and ``solution'' focused processes. In this paper, we study a large dataset of US Senator postings on Twitter (1.68m tweets in total). Our objective is to develop an automated method to label Senatorial posts as either in the problem or solution streams. Two academic policy experts labeled a subset of 3967 tweets as either problem, solution, or other (anything not problem or solution). We split off a subset of 500 tweets into a test set, with the remaining 3467 used for training. During development, this training set was further split by 60/20/20 proportions for fitting, validation, and development test sets. We investigated supervised learning methods for building problem/solution classifiers directly on the training set, evaluating their performance in terms of F1 score on the validation set, allowing us to rapidly iterate through models and hyperparameters, achieving an average weighted F1 score of above 0.8 on cross validation across the three categories using a BERTweet Base model.

Blake VanBerlo, Alexander Wong, Jesse Hoey et al.

Self-supervised learning (SSL) is one strategy for addressing the paucity of labelled data in medical imaging by learning representations from unlabelled images. Contrastive and non-contrastive SSL methods produce learned representations that are similar for pairs of related images. Such pairs are commonly constructed by randomly distorting the same image twice. The videographic nature of ultrasound offers flexibility for defining the similarity relationship between pairs of images. In this study, we investigated the effect of utilizing proximal, distinct images from the same B-mode ultrasound video as pairs for SSL. Additionally, we introduced a sample weighting scheme that increases the weight of closer image pairs and demonstrated how it can be integrated into SSL objectives. Named Intra-Video Positive Pairs (IVPP), the method surpassed previous ultrasound-specific contrastive learning methods' average test accuracy on COVID-19 classification with the POCUS dataset by $\ge 1.3\%$. Detailed investigations of IVPP's hyperparameters revealed that some combinations of IVPP hyperparameters can lead to improved or worsened performance, depending on the downstream task. Guidelines for practitioners were synthesized based on the results, such as the merit of IVPP with task-specific hyperparameters, and the improved performance of contrastive methods for ultrasound compared to non-contrastive counterparts.

Blake VanBerlo, Alexander Wong, Jesse Hoey et al.

Data augmentation is a central component of joint embedding self-supervised learning (SSL). Approaches that work for natural images may not always be effective in medical imaging tasks. This study systematically investigated the impact of data augmentation and preprocessing strategies in SSL for lung ultrasound. Three data augmentation pipelines were assessed: (1) a baseline pipeline commonly used across imaging domains, (2) a novel semantic-preserving pipeline designed for ultrasound, and (3) a distilled set of the most effective transformations from both pipelines. Pretrained models were evaluated on multiple classification tasks: B-line detection, pleural effusion detection, and COVID-19 classification. Experiments revealed that semantics-preserving data augmentation resulted in the greatest performance for COVID-19 classification - a diagnostic task requiring global image context. Cropping-based methods yielded the greatest performance on the B-line and pleural effusion object classification tasks, which require strong local pattern recognition. Lastly, semantics-preserving ultrasound image preprocessing resulted in increased downstream performance for multiple tasks. Guidance regarding data augmentation and preprocessing strategies was synthesized for practitioners working with SSL in ultrasound.

Zahra Sheikhbahaee, Dongshu Luo, Blake VanBerlo et al.

One's ability to learn a generative model of the world without supervision depends on the extent to which one can construct abstract knowledge representations that generalize across experiences. To this end, capturing an accurate statistical structure from observational data provides useful inductive biases that can be transferred to novel environments. Here, we tackle the problem of learning to control dynamical systems by applying Bayesian nonparametric methods, which is applied to solve visual servoing tasks. This is accomplished by first learning a state space representation, then inferring environmental dynamics and improving the policies through imagined future trajectories. Bayesian nonparametric models provide automatic model adaptation, which not only combats underfitting and overfitting, but also allows the model's unbounded dimension to be both flexible and computationally tractable. By employing Gaussian processes to discover latent world dynamics, we mitigate common data efficiency issues observed in reinforcement learning and avoid introducing explicit model bias by describing the system's dynamics. Our algorithm jointly learns a world model and policy by optimizing a variational lower bound of a log-likelihood with respect to the expected free energy minimization objective function. Finally, we compare the performance of our model with the state-of-the-art alternatives for continuous control tasks in simulated environments.

Jerry Huang, Joshua Jung, Neil Budnarain et al.

This paper presents the design of a cooperative multi-player betting game, Trust-ya, as a model of some elements of status processes in human groups. The game is designed to elicit status-driven leader-follower behaviours as a means to observe and influence social hierarchy. It involves a Bach/Stravinsky game of deference in a group, in which people on each turn can either invest with another player or hope someone invests with them. Players who receive investment capital are able to gamble for payoffs from a central pool which then can be shared back with those who invested (but a portion of it may be kept, including all of it). The bigger gambles (people with more investors) get bigger payoffs. Thus, there is a natural tendency for players to coalesce as investors around a 'leader' who gambles, but who also shares sufficiently from their winnings to keep the investors 'hanging on'. The 'leader' will want to keep as much as possible for themselves, however. The game is played anonymously, but a set of 'status symbols' can be purchased which have no value in the game itself, but can serve as a 'cheap talk' communication device with other players. This paper introduces the game, relates it to status theory in social psychology, and shows some simple simulated and human experiments that demonstrate how the game can be used to study status processes and dynamics in human groups.

Kyle Tilbury, Jesse Hoey

Machine learning has the potential to aid in mitigating the human effects of climate change. Previous applications of machine learning to tackle the human effects in climate change include approaches like informing individuals of their carbon footprint and strategies to reduce it. For these methods to be the most effective they must consider relevant social-psychological factors for each individual. Of social-psychological factors at play in climate change, affect has been previously identified as a key element in perceptions and willingness to engage in mitigative behaviours. In this work, we propose an investigation into how affect could be incorporated to enhance machine learning based interventions for climate change. We propose using affective agent-based modelling for climate change as well as the use of a simulated climate change social dilemma to explore the potential benefits of affective machine learning interventions. Behavioural and informational interventions can be a powerful tool in helping humans adopt mitigative behaviours. We expect that utilizing affective ML can make interventions an even more powerful tool and help mitigative behaviours become widely adopted.

Nabiha Asghar, Ivan Kobyzev, Jesse Hoey et al.

State-of-the-art neural dialogue systems excel at syntactic and semantic modelling of language, but often have a hard time establishing emotional alignment with the human interactant during a conversation. In this work, we bring Affect Control Theory (ACT), a socio-mathematical model of emotions for human-human interactions, to the neural dialogue generation setting. ACT makes predictions about how humans respond to emotional stimuli in social situations. Due to this property, ACT and its derivative probabilistic models have been successfully deployed in several applications of Human-Computer Interaction, including empathetic tutoring systems, assistive healthcare devices and two-person social dilemma games. We investigate how ACT can be used to develop affect-aware neural conversational agents, which produce emotionally aligned responses to prompts and take into consideration the affective identities of the interactants.

Aaron W. Li, Veronica Jiang, Steven Y. Feng et al.

For conversational AI and virtual assistants to communicate with humans in a realistic way, they must exhibit human characteristics such as expression of emotion and personality. Current attempts toward constructing human-like dialogue agents have presented significant difficulties. We propose Human Level Attributes (HLAs) based on tropes as the basis of a method for learning dialogue agents that can imitate the personalities of fictional characters. Tropes are characteristics of fictional personalities that are observed recurrently and determined by viewers' impressions. By combining detailed HLA data with dialogue data for specific characters, we present a dataset, HLA-Chat, that models character profiles and gives dialogue agents the ability to learn characters' language styles through their HLAs. We then introduce a three-component system, ALOHA (which stands for Artificial Learning of Human Attributes), that combines character space mapping, character community detection, and language style retrieval to build a character (or personality) specific language model. Our preliminary experiments demonstrate that two variations of ALOHA, combined with our proposed dataset, can outperform baseline models at identifying the correct dialogue responses of chosen target characters, and are stable regardless of the character's identity, the genre of the show, and the context of the dialogue.

Steven Y. Feng, Aaron W. Li, Jesse Hoey

In this paper, we present a novel method for measurably adjusting the semantics of text while preserving its sentiment and fluency, a task we call semantic text exchange. This is useful for text data augmentation and the semantic correction of text generated by chatbots and virtual assistants. We introduce a pipeline called SMERTI that combines entity replacement, similarity masking, and text infilling. We measure our pipeline's success by its Semantic Text Exchange Score (STES): the ability to preserve the original text's sentiment and fluency while adjusting semantic content. We propose to use masking (replacement) rate threshold as an adjustable parameter to control the amount of semantic change in the text. Our experiments demonstrate that SMERTI can outperform baseline models on Yelp reviews, Amazon reviews, and news headlines.

Jesse Hoey, Neil J. MacKinnon

The presence of artificial agents in human social networks is growing. From chatbots to robots, human experience in the developed world is moving towards a socio-technical system in which agents can be technological or biological, with increasingly blurred distinctions between. Given that emotion is a key element of human interaction, enabling artificial agents with the ability to reason about affect is a key stepping stone towards a future in which technological agents and humans can work together. This paper presents work on building intelligent computational agents that integrate both emotion and cognition. These agents are grounded in the well-established social-psychological Bayesian Affect Control Theory (BayesAct). The core idea of BayesAct is that humans are motivated in their social interactions by affective alignment: they strive for their social experiences to be coherent at a deep, emotional level with their sense of identity and general world views as constructed through culturally shared symbols. This affective alignment creates cohesive bonds between group members, and is instrumental for collaborations to solidify as relational group commitments. BayesAct agents are motivated in their social interactions by a combination of affective alignment and decision theoretic reasoning, trading the two off as a function of the uncertainty or unpredictability of the situation. This paper provides a high-level view of dual process theories and advances BayesAct as a plausible, computationally tractable model based in social-psychological theory. We introduce a revised BayesAct model that more deeply integrates social-psychological theorising, and we demonstrate a component of the model as being sufficient to account for cognitive biases about fairness, dissonance and conformity. We show how the model can unify different exploration strategies in reinforcement learning.

Moojan Ghafurian, Neil Budnarain, Jesse Hoey

In this paper, we analyze the performance of an agent developed according to a well-accepted appraisal theory of human emotion with respect to how it modulates play in the context of a social dilemma. We ask if the agent will be capable of generating interactions that are considered to be more human than machine-like. We conduct an experiment with 117 participants and show how participants rate our agent on dimensions of human-uniqueness (which separates humans from animals) and human-nature (which separates humans from machines). We show that our appraisal theoretic agent is perceived to be more human-like than baseline models, by significantly improving both human-nature and human-uniqueness aspects of the intelligent agent. We also show that perception of humanness positively affects enjoyment and cooperation in the social dilemma.

Nabiha Asghar, Pascal Poupart, Jesse Hoey et al.

Existing neural conversational models process natural language primarily on a lexico-syntactic level, thereby ignoring one of the most crucial components of human-to-human dialogue: its affective content. We take a step in this direction by proposing three novel ways to incorporate affective/emotional aspects into long short term memory (LSTM) encoder-decoder neural conversation models: (1) affective word embeddings, which are cognitively engineered, (2) affect-based objective functions that augment the standard cross-entropy loss, and (3) affectively diverse beam search for decoding. Experiments show that these techniques improve the open-domain conversational prowess of encoder-decoder networks by enabling them to produce emotionally rich responses that are more interesting and natural.

Areej Alhothali, Jesse Hoey

In this paper, we propose an extension to graph-based sentiment lexicon induction methods by incorporating distributed and semantic word representations in building the similarity graph to expand a three-dimensional sentiment lexicon. We also implemented and evaluated the label propagation using four different word representations and similarity metrics. Our comprehensive evaluation of the four approaches was performed on a single data set, demonstrating that all four methods can generate a significant number of new sentiment assignments with high accuracy. The highest correlations (tau=0.51) and the lowest error (mean absolute error < 1.1%), obtained by combining both the semantic and the distributional features, outperformed the distributional-based and semantic-based label-propagation models and approached a supervised algorithm.

Shehroz S. Khan, Jesse Hoey

A fall is an abnormal activity that occurs rarely; however, missing to identify falls can have serious health and safety implications on an individual. Due to the rarity of occurrence of falls, there may be insufficient or no training data available for them. Therefore, standard supervised machine learning methods may not be directly applied to handle this problem. In this paper, we present a taxonomy for the study of fall detection from the perspective of availability of fall data. The proposed taxonomy is independent of the type of sensors used and specific feature extraction/selection methods. The taxonomy identifies different categories of classification methods for the study of fall detection based on the availability of their data during training the classifiers. Then, we present a comprehensive literature review within those categories and identify the approach of treating a fall as an abnormal activity to be a plausible research direction. We conclude our paper by discussing several open research problems in the field and pointers for future research.

Shehroz S. Khan, Michelle E. Karg, Dana Kulic et al.

Identification of falls while performing normal activities of daily living (ADL) is important to ensure personal safety and well-being. However, falling is a short term activity that occurs infrequently. This poses a challenge to traditional classification algorithms, because there may be very little training data for falls (or none at all). This paper proposes an approach for the identification of falls using a wearable device in the absence of training data for falls but with plentiful data for normal ADL. We propose three `X-Factor' Hidden Markov Model (XHMMs) approaches. The XHMMs model unseen falls using "inflated" output covariances (observation models). To estimate the inflated covariances, we propose a novel cross validation method to remove "outliers" from the normal ADL that serve as proxies for the unseen falls and allow learning the XHMMs using only normal activities. We tested the proposed XHMM approaches on two activity recognition datasets and show high detection rates for falls in the absence of fall-specific training data. We show that the traditional method of choosing a threshold based on maximum of negative of log-likelihood to identify unseen falls is ill-posed for this problem. We also show that supervised classification methods perform poorly when very limited fall data are available during the training phase.

Hadi Hosseini, Jesse Hoey, Robin Cohen

This paper considers a novel approach to scalable multiagent resource allocation in dynamic settings. We propose an approximate solution in which each resource consumer is represented by an independent MDP-based agent that models expected utility using an average model of its expected access to resources given only limited information about all other agents. A global auction-based mechanism is proposed for allocations based on expected regret. We assume truthful bidding and a cooperative coordination mechanism, as we are considering healthcare scenarios. We illustrate the performance of our coordinated MDP approach against a Monte-Carlo based planning algorithm intended for large-scale applications, as well as other approaches suitable for allocating medical resources. The evaluations show that the global utility value across all consumer agents is closer to optimal when using our algorithms under certain time constraints, with low computational cost. As such, we offer a promising approach for addressing complex resource allocation problems that arise in healthcare settings.

Jesse Hoey, Tobias Schroeder, Areej Alhothali

This paper describes a novel method for building affectively intelligent human-interactive agents. The method is based on a key sociological insight that has been developed and extensively verified over the last twenty years, but has yet to make an impact in artificial intelligence. The insight is that resource bounded humans will, by default, act to maintain affective consistency. Humans have culturally shared fundamental affective sentiments about identities, behaviours, and objects, and they act so that the transient affective sentiments created during interactions confirm the fundamental sentiments. Humans seek and create situations that confirm or are consistent with, and avoid and supress situations that disconfirm or are inconsistent with, their culturally shared affective sentiments. This "affect control principle" has been shown to be a powerful predictor of human behaviour. In this paper, we present a probabilistic and decision-theoretic generalisation of this principle, and we demonstrate how it can be leveraged to build affectively intelligent artificial agents. The new model, called BayesAct, can maintain multiple hypotheses about sentiments simultaneously as a probability distribution, and can make use of an explicit utility function to make value-directed action choices. This allows the model to generate affectively intelligent interactions with people by learning about their identity, predicting their behaviours using the affect control principle, and taking actions that are simultaneously goal-directed and affect-sensitive. We demonstrate this generalisation with a set of simulations. We then show how our model can be used as an emotional "plug-in" for artificially intelligent systems that interact with humans in two different settings: an exam practice assistant (tutor) and an assistive device for persons with a cognitive disability.

Jesse Hoey, Robert St-Aubin, Alan Hu et al.

Markov decisions processes (MDPs) are becoming increasing popular as models of decision theoretic planning. While traditional dynamic programming methods perform well for problems with small state spaces, structured methods are needed for large problems. We propose and examine a value iteration algorithm for MDPs that uses algebraic decision diagrams(ADDs) to represent value functions and policies. An MDP is represented using Bayesian networks and ADDs and dynamic programming is applied directly to these ADDs. We demonstrate our method on large MDPs (up to 63 million states) and show that significant gains can be had when compared to tree-structured representations (with up to a thirty-fold reduction in the number of nodes required to represent optimal value functions).

Marek Grzes, Jesse Hoey, Shehroz Khan et al.

Assistive systems for persons with cognitive disabilities (e.g. dementia) are difficult to build due to the wide range of different approaches people can take to accomplishing the same task, and the significant uncertainties that arise from both the unpredictability of client's behaviours and from noise in sensor readings. Partially observable Markov decision process (POMDP) models have been used successfully as the reasoning engine behind such assistive systems for small multi-step tasks such as hand washing. POMDP models are a powerful, yet flexible framework for modelling assistance that can deal with uncertainty and utility. Unfortunately, POMDPs usually require a very labour intensive, manual procedure for their definition and construction. Our previous work has described a knowledge driven method for automatically generating POMDP activity recognition and context sensitive prompting systems for complex tasks. We call the resulting POMDP a SNAP (SyNdetic Assistance Process). The spreadsheet-like result of the analysis does not correspond to the POMDP model directly and the translation to a formal POMDP representation is required. To date, this translation had to be performed manually by a trained POMDP expert. In this paper, we formalise and automate this translation process using a probabilistic relational model (PRM) encoded in a relational database. We demonstrate the method by eliciting three assistance tasks from non-experts. We validate the resulting POMDP models using case-based simulations to show that they are reasonable for the domains. We also show a complete case study of a designer specifying one database, including an evaluation in a real-life experiment with a human actor.