Auditing and Debugging Deep Learning Models via Flip Points: Individual-Level and Group-Level Analysis

Deep learning models have been criticized for their lack of easy interpretation, which undermines confidence in their use for important applications. Nevertheless, they are consistently utilized in many applications, consequential to humans’ lives, usually because of their better performance. Therefore, there is a great need for computational methods that can explain, audit, and debug such models. Here, we use flip points to accomplish these goals for deep learning classifiers used in social applications. A trained deep learning classifier is a mathematical function that maps inputs to classes. By way of training, the function partitions its domain and assigns a class to each of the partitions. Partitions are defined by the decision boundaries which are expected to be geometrically complex. This complexity is usually what makes deep learning models powerful classifiers. Flip points are points on those boundaries and, therefore, the key to understanding and changing the functional behavior of models. We use advanced numerical optimization techniques and state-of-the-art methods in numerical linear algebra, such as rank determination and reduced-order models to compute and analyze them. The resulting insight into the decision boundaries of a deep model can clearly explain the model’s output on the individual level, via an explanation report that is understandable by non-experts. We also develop a procedure to understand and audit model behavior towards groups of people. We show that examining decision boundaries of models in certain subspaces can reveal hidden biases that are not easily detectable. Flip points can also be used as synthetic data to alter the decision boundaries of a model and improve their functional behaviors. We demonstrate our methods by investigating several models trained on standard datasets used in social applications of machine learning. We also identify the features that are most responsible for particular classifications and misclassifications. Finally, we discuss the implications of our auditing procedure in the public policy domain.

An Innovative Project of a Bionic Robot for Social Applications in Felinotherapy

The paper describes an innovative design of a bionic robot for applications in felinotherapy supporting hospital and home psychotherapeutic treatment of bedridden children and adults. The project was engineered by biomimicrating a biological cat, reaching its robotic model. Particular attention in this process was devoted to capturing the essence of feline motorics behavior and the possibility of mapping them in a mechatronic model. The geometry, kinematics and kinetics of this model were analyzed, creating assumptions for its practical implementation in the real mechanism of cat skeleton movement. The used software used the topology of elements in Autodesk Fusion 360 Simulation workspace by performing the critical elements of the mechatronic model in print using SLS technology. The work was also supported by a graphical simulation in the PyBullet environment.

Impact of Mobile Applications in Changing the Tourist Experience

This exploratory study investigates the impact of technology on the tourist experience. This research analyzed the effect of new technologies on the behaviour of new tourism consumers, the importance of mobile technology in the tourism industry, and the phenomenon of mobile tourism in the change of the tourism experience. It also studied the spillover effect that these changes have had on the use of mobile applications in tourism, trying to understand their impact on the tourism experience. A quantitative and qualitative analysis supported this investigation through a survey of 110 Portuguese tourists using applications and two interviews with administrators of mobile application development companies, and, finally, through the analysis of four case studies. The main conclusion is that smartphones and mobile travel applications can substantially alter the tourist experience. Tourists assume that smartphones have motivated changes in the activities and emotions experienced, especially in social applications that allow them to maintain contact with their friends and families and increase security levels.

Serious information in hedonic social applications: affordances, self-determination and health information adoption in TikTok

PurposeHedonic social applications have been increasingly popular among health information consumers. However, it remains unclear what motivates consumers to adopt health information in hedonic applications when they have alternative choices of more formal health information sources. Building on the self-determination theory and the affordances lens, this study aims to investigate how different affordances on hedonic social applications affect consumers' basic psychological needs and further influence their intention to adopt health information on such applications.Design/methodology/approachAs TikTok demonstrated great potential in disseminating health information, we developed a model that we analyze using the PLS-SEM technique with data collected from a valid research sample of 384 respondents with health information seeking or encountering experience in TikTok.FindingsThe results suggested that health information adoption in hedonic social applications is significantly predicted by the satisfaction of consumers' basic psychological needs, namely autonomy, relatedness and competence. Moreover, the satisfaction of basic psychological needs is positively affected by affordances provided by the hedonic social applications. The hedonic affordances positively influence autonomy satisfaction, while the connective affordances positively affect relatedness satisfaction, and the utilitarian affordances positively support competence satisfaction.Originality/valueThe study indicates that hedonic social applications such as TikTok could be an important channel for consumers to access and adopt health information. The study contributes to the literature by proposing a theoretical model that explains consumers' health information adoption and yields practical implications for designers and service providers of hedonic social applications.

A Survey on Classical and Modern Face Recognition Techniques

During the previous few centuries, facial recognition systems have become a popular research topic. On account of its extraordinary success and vast social applications; it has attracted significant study attention from a wide range of disciplines in the last five years - including “computer-vision”, “artificial-intelligence”, and “machine-learning”. As with most face recognition systems, the fundamental goal involves recognizing a person's identity by means of images, video, data streams, and context information. As a result of our research; we've outlined some of the most important applications, difficulties, and trends in scientific and social domains. This research, the primary goal is to summarize modern facial recognition algorithms and to gain a general perceptive of how these techniques act on diverse datasets. Aside from that, we also explore some significant problems like illumination variation, position, aging, occlusion, cosmetics, scale, and background are some of the primary challenges we examine. In addition to traditional face recognition approaches, the most recent research topics such as sparse models, deep learning, and fuzzy set theory are examined in depth. There's also a quick discussion of basic techniques, as well as a more in-depth. As a final point, this research explores the future of facial recognition technologies and their possible importance in the emerging digital society.

A Survey on Different Recommendation Techniques

Recommender systems are used to provide recommendation or suggestions on services and items to the users. It provides suggestions which based on prediction. Prediction plays an important role in recommender systems while making the recommendation for users, prediction ensures the quality of suggestions for their users. Recommendation system used in various applications such as e-commerce websites, social Applications, search engines, or for daily decisions that we need to make. To generate prediction for recommender systems many techniques have been proposed. The goal of the paper is to provide a general overview on various recommendation techniques. It covers the working of different recommendation techniques along with its classification and also their advantages and disadvantages.

User Intention of Anonymous Social Application “Soul” in China: Analysis based on an Extended Technology Acceptance Model

Anonymity is an inherent attribute of the Internet. Depending on pseudonyms, cyber citizens can role play and present themselves by reconstructing a different identity. In order to satisfy the needs of anonymous self-expression, anonymous social applications have become popular worldwide. In this paper, we conduct a survey regarding user intention (UI) of “Soul”, which is a popular anonymous social media application in China, especially for the youth. For this purpose, we design an adapted technology acceptance model (TAM) consisting of seven influencing factors, i.e., perceived usefulness (PU), perceived ease of use (PEOU), perceived anonymity (PA), perceived privacy riskiness (PPR), subjective norms (SN), emotional attachments (EA) and perceived interactivity (PI). Both the measurement and structural models are tested via partial least squares structural equation model. The results show that PU, PEOU, PPR and PI have a significant relationship with UI. Therein, both SN and EA can impact PU, and meanwhile, the direct paths between PI → PEOU, PA → PPR also exist. Contrary to expectation, the effect of SN on UI is not directly significant. The proposed model is able to explain 64.1% of variance for UI among Soul users. The results suggest that the proposed constructs provide relatively good explanations for the continuous intention to use the Soul app.

Remote support to victims of violence against women and domestic violence during the COVID-19 pandemic

Purpose This paper aims to characterize the type of support provided to victims of violence against women and domestic violence (VAWDV) during the first lockdown, assessing the training of professionals to use remote support (RS). Design/methodology/approach This cross-sectional study involves a sample of 196 support professionals, mainly women (91.8%) and who integrate the Portuguese National Support Network for victims of domestic violence (NSNVDV) (Mean age = 36.49; SD = 10.52). Findings Telephone emerges as the main RS communication media used in the lockdown (43.9%) and the emergency state periods (57.1%). Participants reported to have never used any social applications (41.8% vs 41.8%) or videoconference (46.4% vs 58.2%), in both periods assessed, i.e. lockdown and emergency state, respectively, and 82.7% assumed to have no training with RS to assist VAWDV victims. However, support professionals recognized several advantages in using RS such as dealing with isolation, reducing inhibition, fear and shame and in promoting the victims’ empowerment. Research limitations/implications Given the exploratory nature of this study, only descriptive analyzes were conducted. Originality/value During the COVID-19 pandemic, little is known about effective RS given by professionals to victims of VAWDV in the Portuguese context. The paper aims to add knowledge to the studied field.