Complex Real-world Settings Research Articles

In-Depth Analysis of GAF-Net: Comparative Fusion Approaches in Video-Based Person Re-Identification

This study provides an in-depth analysis of GAF-Net, a novel model for video-based person re-identification (Re-ID) that matches individuals across different video sequences. GAF-Net combines appearance-based features with gait-based features derived from skeletal data, offering a new approach that diverges from traditional silhouette-based methods. We thoroughly examine each module of GAF-Net and explore various fusion methods at the both score and feature levels, extending beyond initial simple concatenation. Comprehensive evaluations on the iLIDS-VID and MARS datasets demonstrate GAF-Net’s effectiveness across scenarios. GAF-Net achieves state-of-the-art 93.2% rank-1 accuracy on iLIDS-VID’s long sequences, while MARS results (86.09% mAP, 89.78% rank-1) reveal challenges with shorter, variable sequences in complex real-world settings. We demonstrate that integrating skeleton-based gait features consistently improves Re-ID performance, particularly with long, more informative sequences. This research provides crucial insights into multi-modal feature integration in Re-ID tasks, laying a foundation for the advancement of multi-modal biometric systems for diverse computer vision applications.

Automatic detection of epilepsy from EEGs using a temporal convolutional network with a self-attention layer

BackgroundOver 60% of epilepsy patients globally are children, whose early diagnosis and treatment are critical for their development and can substantially reduce the disease’s burden on both families and society. Numerous algorithms for automated epilepsy detection from EEGs have been proposed. Yet, the occurrence of epileptic seizures during an EEG exam cannot always be guaranteed in clinical practice. Models that exclusively use seizure EEGs for detection risk artificially enhanced performance metrics. Therefore, there is a pressing need for a universally applicable model that can perform automatic epilepsy detection in a variety of complex real-world scenarios.MethodTo address this problem, we have devised a novel technique employing a temporal convolutional neural network with self-attention (TCN-SA). Our model comprises two primary components: a TCN for extracting time-variant features from EEG signals, followed by a self-attention (SA) layer that assigns importance to these features. By focusing on key features, our model achieves heightened classification accuracy for epilepsy detection.ResultsThe efficacy of our model was validated on a pediatric epilepsy dataset we collected and on the Bonn dataset, attaining accuracies of 95.50% on our dataset, and 97.37% (A v. E), and 93.50% (B vs E), respectively. When compared with other deep learning architectures (temporal convolutional neural network, self-attention network, and standardized convolutional neural network) using the same datasets, our TCN-SA model demonstrated superior performance in the automated detection of epilepsy.ConclusionThe proven effectiveness of the TCN-SA approach substantiates its potential as a valuable tool for the automated detection of epilepsy, offering significant benefits in diverse and complex real-world clinical settings.

Communication partner training for student health and social care professionals engaging with people with stroke acquired communication difficulties: A protocol for a realist review.

Background Stroke acquired communication impairments impede effective communication. Consequently, in stroke care, communicative interactions can be challenging for both patients and staff and can predispose patients to increased risk of preventable adverse events. Communication partner training (CPT) can mitigate such negative outcomes by optimising communicative interactions. Providing CPT to student health and social care professionals (SH&SCPs) has the potential to enhance their clinical expertise and experiences and enhance the future clinical care of patients with stroke acquired communication impairments. This research aims to expand our understanding of how CPT is operationalised for SH&SCPs in higher education institutions and determine: what works; for whom; in what contexts; how and why? Methods This review is Phase 1 of a research project employing a realist approach with public and patient involvement (PPI). It incorporates five iterative steps: 1.) Clarifying the scope; 2.) Searching for evidence; 3.) Selecting and appraising evidence; 4.) Data extraction; 5.) Synthesising data and developing a middle range theory explaining how CPT is expected to work for SH&SCPs. An advisory panel, including PPI advisors, content advisors, student advisors, realist advisors and educationalist advisor has been set up to consult throughout the review and collaboratively agree the middle range theory. Discussion While there is an evolving evidence base for CPT, including stroke specific CPT for SH&SCPs, it is acknowledged that there are challenges to its implementation in complex real-world settings. In combining empirical evidence with theoretical understanding, realist review permits synthesis of data from diverse sources and goes beyond determining efficacy to explore generative causation and solutions for real world practice. A middle range realist programme theory that coherently explains how CPT is expected to work when teaching SH&SCPs to communicate with people with stroke acquired communication impairments will provide educators with new insights into CPT development and implementation in their higher education institutions.

Local machine learning model-based multi-objective optimization for managing system interdependencies in production: A case study from the ironmaking industry

Modeling interdependencies in a production process is a vital aspect of process engineering. Being built on the fundamental understanding of the process and capturing underlying physical and chemical mechanisms, this task is traditionally carried out through first-principle modeling. However, due to various assumptions and parameter approximation, such approach sometimes struggles to accurately model process dynamics. This is partially due to the requirements for process stability in mission-critical scenarios, where it is not viable to conduct extensive experiments that would enable the development of comprehensive models, and hence, the process control is still partially dependent on the experience of the process operator. In this paper, a data-driven approach based on production data is introduced to address the challenge of discovering, understanding, and modeling interdependencies in a production process by constructing a constrained multi-objective optimization problem. The applicability of the proposed approach is demonstrated through a case study in the ironmaking industry, whereby a set of Machine Learning and causality-based methods is provided while highlighting the significance of transparency and use of domain knowledge in the development of data-driven models. Based on the real data from a Sintering plant, interdependencies are quantified between five key process parameters, which, when supplemented with other control parameters, result in a higher overall process output leading to reductions in operational costs and environmental impact. Such approach shows potential for applications in a broader context across other fields of science and engineering, where it can be used to improve the discovery of interdependencies in complex real-world industrial settings.

Communication partner training for student health and social care professionals engaging with people with stroke acquired communication difficulties: A protocol for a realist review.

Stroke acquired communication impairments impede effective communication. Consequently, in stroke care, communicative interactions can be challenging for both patients and staff and can predispose patients to increased risk of preventable adverse events. Communication partner training (CPT) can mitigate such negative outcomes by optimising communicative interactions. Providing CPT to student health and social care professionals (SH&SCPs) has the potential to enhance their clinical expertise and experiences and enhance the future clinical care of patients with stroke acquired communication impairments. This research aims to expand our understanding of how CPT is operationalised for SH&SCPs in higher education institutions and determine: what works; for whom; in what contexts; how and why? This review is Phase 1 of a research project employing a realist approach with public and patient involvement (PPI). It incorporates five iterative steps: 1.) Clarifying the scope; 2.) Searching for evidence; 3.) Selecting and appraising evidence; 4.) Data extraction; 5.) Synthesising data and developing a middle range theory explaining how CPT is expected to work for SH&SCPs. An advisory panel, including PPI advisors, content advisors, student advisors, realist advisors and educationalist advisor has been set up to consult throughout the review and collaboratively agree the middle range theory. While there is an evolving evidence base for CPT, including stroke specific CPT for SH&SCPs, it is acknowledged that there are challenges to its implementation in complex real-world settings. In combining empirical evidence with theoretical understanding, realist review permits synthesis of data from diverse sources and goes beyond determining efficacy to explore generative causation and solutions for real world practice. A middle range realist programme theory that coherently explains how CPT is expected to work when teaching SH&SCPs to communicate with people with stroke acquired communication impairments will provide educators with new insights into CPT development and implementation in their higher education institutions.

Collaborative Perception-The Missing Piece in Realizing Fully Autonomous Driving.

Environment perception plays a crucial role in enabling collaborative driving automation, which is considered to be the ground-breaking solution to tackling the safety, mobility, and sustainability challenges of contemporary transportation systems. Despite the fact that computer vision for object perception is undergoing an extraordinary evolution, single-vehicle systems' constrained receptive fields and inherent physical occlusion make it difficult for state-of-the-art perception techniques to cope with complex real-world traffic settings. Collaborative perception (CP) based on various geographically separated perception nodes was developed to break the perception bottleneck for driving automation. CP leverages vehicle-to-vehicle and vehicle-to-infrastructure communication to enable vehicles and infrastructure to combine and share information to comprehend the surrounding environment beyond the line of sight and field of view to enhance perception accuracy, lower latency, and remove perception blind spots. In this article, we highlight the need for an evolved version of the collaborative perception that should address the challenges hindering the realization of level 5 AD use cases by comprehensively studying the transition from classical perception to collaborative perception. In particular, we discuss and review perception creation at two different levels: vehicle and infrastructure. Furthermore, we also study the communication technologies and three different collaborative perception message-sharing models, their comparison analyzing the trade-off between the accuracy of the transmitted data and the communication bandwidth used for data transmission, and the challenges therein. Finally, we discuss a range of crucial challenges and future directions of collaborative perception that need to be addressed before a higher level of autonomy hits the roads.

Providing Education to the 1st Year Students at the Medical University in the Conditions of Martial State in Ukraine

Learning is a complex creative process based on the use of modern technologies and taking into account the psycho-physiological properties of an individual. The martial law introduced significant changes in the operation mode of higher education institutions in Ukraine. Therefore, the problem of organization and implementation training in higher education institutions is very relevant in current conditions. The purpose is the study and application of those modern forms and methods of education that would provide a sufficient level of knowledge in martial law, as well as the study of the peculiarities of student’s motivation in martial law. Methodology. We conducted a comprehensive analysis of sources about various teaching methods and generalization own experience of teaching Medical Biology in online format at Zaporizhzhia State Medical University. Results. It was implemented the interaction concept of higher education students with the teaching staff and educational activity was monitored, Zaporizhzhia State Medical University introduced a remote form of education using MS Teams and the edX platform. Based on Teams, the work was carried out to create and fill in online platforms with an educational material, a content visualization with a large amount of illustrative materials, video materials, packages of test tasks in MS Forms, along with educational methods adaptation, including interactive ones, to a distance form. Interactive distance learning means are very diverse. They include synchronous tools such as video conferencing and virtual classroom discussions, as well as asynchronous tools such as online forums and chats. Interactive distance learning means are crucial for several reasons: increasing contact, improving communication and collaboration, and contributing to a more dynamic and collaborative learning environment. While our teachers in real-time use synchronous interactive methods, asynchronous methods provide great flexibility for students regarding when and where they can participate in classes, that is especially important for students in complex real-world settings of martial law. Conclusions: Distance learning in modern higher education has become a forced alternative to traditional forms despite the difficulties ensuring the educational process under challenging conditions. Interactive means are becoming increasingly important in distance learning, as they help boost student’s communication and interaction, improve collaboration, and provide greater flexibility and convenience in the distance learning process for students and teachers. Modern technologies and means will be appropriate if we take into account the students’ psycho-physiological state and the degree of their motivation.

Mobile cognition: imaging the human brain in the 'real world'.

Cognitive neuroscience studies in humans have enabled decades of impactful discoveries but have primarily been limited to recording the brain activity of immobile participants in a laboratory setting. In recent years, advances in neuroimaging technologies have enabled recordings of human brain activity to be obtained during freely moving behaviours in the real world. Here, we propose that these mobile neuroimaging methods can provide unique insights into the neural mechanisms of human cognition and contribute to the development of novel treatments for neurological and psychiatric disorders. We further discuss the challenges associated with studying naturalistic human behaviours in complex real-world settings as well as strategies for overcoming them. We conclude that mobile neuroimaging methods have the potential to bring about a new era of cognitive neuroscience in which neural mechanisms can be studied with increased ecological validity and with the ability to address questions about natural behaviour and cognitive processes in humans engaged in dynamic real-world experiences.

Point estimate observers: A new class of models for perceptual decision making.

Bayesian optimal inference is often heralded as a principled, general framework for human perception. However, optimal inference requires integration over all possible world states, which quickly becomes intractable in complex real-world settings. Additionally, deviations from optimal inference have been observed in human decisions. A number of approximation methods have previously been suggested, such as sampling methods. In this study, we additionally propose point estimate observers, which evaluate only a single best estimate of the world state per response category. We compare the predicted behavior of these model observers to human decisions in five perceptual categorization tasks. Compared to the Bayesian observer, the point estimate observer loses decisively in one task, ties in two and wins in two tasks. Two sampling observers also improve upon the Bayesian observer, but in a different set of tasks. Thus, none of the existing general observer models appears to fit human perceptual decisions in all situations, but the point estimate observer is competitive with other observer models and may provide another stepping stone for future model development. (PsycInfo Database Record (c) 2023 APA, all rights reserved).

New municipalism and the governance of urban transitions to sustainability

Cities play increasingly recognised roles in global climate change responses: as change laboratories, spaces of opportunity, and as administrative and economic hubs that concentrate human and financial resources and needs. They host high climate mitigation potential and acute climate adaptation vulnerabilities. Scholarship flags conventional urban planning approaches to limit global warming to 1.5°C as inadequate. Yet urban sustainability transitions literature features few examples of functioning alternative governance and planning paradigms. This paper assesses one such approach, new municipalism: social movements centred on a democratic transformation of the local economy and state. We combine attention to urban sustainability transitions and new municipalism research to interrogate whether and how the latter can facilitate the provision of leadership and institutional arrangements that enable urban transformation to sustainability. Our desk study considers two prominent examples of new municipalism in Spain, where Barcelona en Comú and Ahora Madrid arose as anti-austerity movements to combat neoliberal urban agendas during the 2010s. We find that the praxis of collective decision-making associated with new municipalism does offer inclusive, innovative policy pathways and the potential to implement experimental knowledge and learning in complex real-world settings at the urban scale. We argue, however, that powerful neoliberal mechanisms impose structural constraints on the very push for deep political change that new municipalist movements embody. By linking transformative climate governance needs with new municipalism movements and wider political economic structuring forces, we explicate the tensions and contested dynamics of institutionalising progressive social movements in the multi-scalar governance of urban sustainability transformation.

Stressors Increase the Impacts of Coastal Macrofauna Biodiversity Loss on Ecosystem Multifunctionality

There is substantial evidence that biodiversity underpins ecosystem functioning, but it is unclear how these relationships change with multiple stressors in complex real-world settings. Coastal zones are affected by numerous stressors (for example, sediment input and nutrient runoff from land) and the cumulative effects of these stressors may result in pronounced and unexpected changes in the functioning of ecosystems. To investigate the cumulative effects of turbidity and elevated nutrients on coastal biodiversity-ecosystem functioning relationships, we performed a large-scale field experiment manipulating in situ sediment porewater ammonium concentrations and measured multiple ecosystem functions related to carbon fixation and mineralisation in 15 estuaries with varying levels of turbidity. The results indicated that the benthic macrofauna diversity (species richness, abundance, and functional richness) declined with increased porewater ammonium concentrations and there were clear thresholds in light levels at the seafloor in relation to the biodiversity-ecosystem function relationships. Multifunctionality indices (an integrated index of all measured functions) in moderately turbid and turbid estuaries (daily mean seafloor PAR < 420 µmol m−2 s−1) decreased with the loss of macrofauna biodiversity. Functioning in low-turbidity estuaries (daily mean PAR > 420 µmol m−2 s−1) however remained relatively constant, suggesting that they were more resilient against the nutrient-induced biodiversity loss. Our results demonstrate that ecosystems already stressed by stressors that alter functional performance (turbidity) may be more prone to loss of overall functioning if biodiversity is reduced by another stressor (nutrient enrichment), highlighting the potential snowballing effects of cumulative change.

Scaling up our understanding of tipping points.

Anthropogenic activities are increasingly affecting ecosystems across the globe. Meanwhile, empirical and theoretical evidence suggest that natural systems can exhibit abrupt collapses in response to incremental increases in the stressors, sometimes with dramatic ecological and economic consequences. These catastrophic shifts are faster and larger than expected from the changes in the stressors and happen once a tipping point is crossed. The primary mechanisms that drive ecosystem responses to perturbations lie in their architecture of relationships, i.e. how species interact with each other and with the physical environment and the spatial structure of the environment. Nonetheless, existing theoretical work on catastrophic shifts has so far largely focused on relatively simple systems that have either few species and/or no spatial structure. This work has laid a critical foundation for understanding how abrupt responses to incremental stressors are possible, but it remains difficult to predict (let alone manage) where or when they are most likely to occur in more complex real-world settings. Here, we discuss how scaling up our investigations of catastrophic shifts from simple to more complex-species rich and spatially structured-systems could contribute to expanding our understanding of how nature works and improve our ability to anticipate the effects of global change on ecological systems. This article is part of the theme issue 'Ecological complexity and the biosphere: the next 30 years'.

MobilePhys

Camera-based contactless photoplethysmography refers to a set of popular techniques for contactless physiological measurement. The current state-of-the-art neural models are typically trained in a supervised manner using videos accompanied by gold standard physiological measurements. However, they often generalize poorly out-of-domain examples (i.e., videos that are unlike those in the training set). Personalizing models can help improve model generalizability, but many personalization techniques still require some gold standard data. To help alleviate this dependency, in this paper, we present a novel mobile sensing system called MobilePhys, the first mobile personalized remote physiological sensing system, that leverages both front and rear cameras on a smartphone to generate high-quality self-supervised labels for training personalized contactless camera-based PPG models. To evaluate the robustness of MobilePhys, we conducted a user study with 39 participants who completed a set of tasks under different mobile devices, lighting conditions/intensities, motion tasks, and skin types. Our results show that MobilePhys significantly outperforms the state-of-the-art on-device supervised training and few-shot adaptation methods. Through extensive user studies, we further examine how does MobilePhys perform in complex real-world settings. We envision that calibrated or personalized camera-based contactless PPG models generated from our proposed dual-camera mobile sensing system will open the door for numerous future applications such as smart mirrors, fitness and mobile health applications.

Causal Judgment in the Wild: Evidence from the 2020 U.S. Presidential Election.

When explaining why an event occurred, people intuitively highlight some causes while ignoring others. How do people decide which causes to select? Models of causal judgment have been evaluated in simple and controlled laboratory experiments, but they have yet to be tested in a complex real-world setting. Here, we provide such a test, in the context of the 2020 U.S. presidential election. Across tens of thousands of simulations of possible election outcomes, we computed, for each state, an adjusted measure of the correlation between a Biden victory in that state and a Biden election victory. These effect size measures accurately predicted the extent to which U.S. participants (N = 207, preregistered) viewed victory in a given state as having caused Biden to win the presidency. Our findings support the theory that people intuitively select as causes of an outcome the factors with the largest standardized causal effect on that outcome across possible counterfactual worlds.

Avoiding Negative Side Effects Due to Incomplete Knowledge of AI Systems

Autonomous agents acting in the real-world often operate based on models that ignore certain aspects of the environment. The incompleteness of any given model – handcrafted or machine acquired – is inevitable due to practical limitations of any modeling technique for complex real-world settings. Due to the limited fidelity of its model, an agent’s actions may have unexpected, undesirable consequences during execution. Learning to recognize and avoid such negative side effects (NSEs) of an agent’s actions is critical to improve the safety and reliability of autonomous systems. Mitigating NSEs is an emerging research topic that is attracting increased attention due to the rapid growth in the deployment of AI systems and their broad societal impacts. This article provides a comprehensive overview of different forms of NSEs and the recent research efforts to address them. We identify key characteristics of NSEs, highlight the challenges in avoiding NSEs, and discuss recently developed approaches, contrasting their benefits and limitations. The article concludes with a discussion of open questions and suggestions for future research directions.

Preparing Health Profession Students for Practice in Complex Real World Settings: How Do Educators Respond to a Model of Capability?

Undergraduate education often leaves future healthcare professionals insufficiently prepared for the complex and unpredictable workplace. Competency-based medical education (CBME) is widely utilised in health professions’ education, but a growing literature highlights its limitations. Capability is a complex concept which builds on competence, while embedding the integration and adaptation of knowledge, skills and personal qualities. While capability can prepare learners for uncertainty and complexity, the concept has been slow to progress within health professions’ education. We introduced the concept of capability to over 110 international health profession educators through six conference workshops, using a visual model to facilitate understanding. Participants’ post-workshop qualitative feedback was collated and thematically analysed. One hundred and thirty-three free text comments were received and four main themes identified: conceptualising capability, capability as curriculum goal, educating for capability, challenges to implementation. The need for capability resonated with participants, although most were unfamiliar with the concept prior to the workshop. The workshop enabled participants to recognise the limitations of CBME’s ‘tick box’ culture when preparing learners to address complexity. The model helped educators grasp the concept of capability and its potential value in health professions’ education. Participants were then able to identify challenges and suggest approaches for implementing capability in practice. Educating for capability is fundamental for preparing students successfully for the ‘real world’ of professional practice. Most participants agreed it should be the end goal of healthcare education curricula. Healthcare educators are likely, however, to need support to embrace this paradigm shift.

Implementation of a Transitional Care Model for Stroke: Perspectives From Frontline Clinicians, Administrators, and COMPASS-TC Implementation Staff.

Stroke is a chronic, complex condition that disproportionally affects older adults. Health systems are evaluating innovative transitional care (TC) models to improve outcomes in these patients. The Comprehensive Post-Acute Stroke Services (COMPASS) Study, a large cluster-randomized pragmatic trial, tested a TC model for patients with stroke or transient ischemic attack discharged home from the hospital. The implementation of COMPASS-TC in complex real-world settings was evaluated to identify successes and challenges with integration into the clinical workflow. We conducted a concurrent process evaluation of COMPASS-TC implementation during the first year of the trial. Qualitative data were collected from 4 sources across 19 intervention hospitals. We analyzed transcripts from 43 conference calls with hospital clinicians, individual and group interviews with leaders and clinicians from 9 hospitals, and 2 interviews with the COMPASS-TC Director of Implementation using iterative thematic analysis. Themes were compared to the domains of the RE-AIM framework. Organizational, individual, and community factors related to Reach, Adoption, and Implementation were identified. Organizational readiness was an additional key factor to successful implementation, in that hospitals that were not "organizationally ready" had more difficulty addressing implementation challenges. Multifaceted TC models are challenging to implement. Facilitators of implementation were organizational commitment and capacity, prioritizing implementation of innovative delivery models to provide comprehensive care, being able to address challenges quickly, implementing systems for tracking patients throughout the intervention, providing clinicians with autonomy and support to address challenges, and adequately resourcing the intervention. NCT02588664.

Absence of DOA Effect but No Proper Test of the Lumberjack Effect: A Reply to Jamieson and Skraaning (2019).

The aim was to evaluate the relevance of the critique offered by Jamieson and Skraaning (2019) regarding the applicability of the lumberjack effect of human-automation interaction to complex real-world settings. The lumberjack effect, based upon a meta-analysis, identifies the consequences of a higher degree of automation-to improve performance and reduce workload-when automation functions as intended, but to degrade performance more, as mediated by a loss of situation awareness (SA) when automation fails. Jamieson and Skraaning provide data from a process control scenario that they assert contradicts the effect. We analyzed key aspects of their simulation, measures, and results which we argue limit the strength of their conclusion that the lumberjack effect is not applicable to complex real-world systems. Our analysis revealed limits in their inappropriate choice of automation, the lack of a routine performance measure, support for the lumberjack effect that was actually provided by subjective measures of the operators, an inappropriate assessment of SA, and a possible limitation of statistical power. We regard these limitations as reasons to temper the strong conclusions drawn by the authors, of no applicability of the lumberjack effect to complex environments. Their findings should be used as an impetus for conducting further research on human-automation interaction in these domains. The collective findings of both Jamieson and Skraaning and our study are applicable to system designers and users in deciding upon the appropriate level of automation to deploy.

Seasonally Varying Secondary Organic Aerosol Formation From In-Situ Oxidation of Near-Highway Air.

The extent to which motor vehicles contribute to ambient secondary organic aerosol (SOA) remains uncertain. Here, we present in situ measurements of SOA formation at a near-highway site with substantial tree-cover 10 m from Interstate 40 near Durham, North Carolina. In July 2015 (summer) and February 2016 (winter), we exposed ambient air to a range of oxidant (O3 and OH) concentrations in an oxidation flow reactor (OFR), resulting in hours to weeks of equivalent atmospheric aging. We observed substantial seasonal variation in SOA formation upon OFR aging; diurnally varying OA enhancements of ∼3-8 μg m-3 were observed in summer and significantly lower enhancements (∼0.5-1 μg m-3) in winter. Measurements in both seasons showed consistent changes in bulk OA properties (chemical composition; volatility) with OFR aging. Mild increases in traffic-related SOA precursors during summer partly explains the seasonal variation. However, biogenic emissions, with sharp temperature dependence, appear to dominate summer OFR-SOA. Our analysis indicates that SOA observed in the OFR is similar (within a factor of 2) to that predicted to form from traffic and biogenic precursors using literature yields, especially in winter. This study highlights the utility of the OFR for studying the prevalence of SOA precursors in complex real-world settings.

Complex Sequential Question Answering: Towards Learning to Converse Over Linked Question Answer Pairs with a Knowledge Graph

While conversing with chatbots, humans typically tend to ask many questions, a significant portion of which can be answered by referring to large-scale knowledge graphs (KG). While Question Answering (QA) and dialog systems have been studied independently, there is a need to study them closely to evaluate such real-world scenarios faced by bots involving both these tasks. Towards this end, we introduce the task of Complex Sequential QA which combines the two tasks of (i) answering factual questions through complex inferencing over a realistic-sized KG of millions of entities, and (ii) learning to converse through a series of coherently linked QA pairs. Through a labor intensive semi-automatic process, involving in-house and crowdsourced workers, we created a dataset containing around 200K dialogs with a total of 1.6M turns. Further, unlike existing large scale QA datasets which contain simple questions that can be answered from a single tuple, the questions in our dialogs require a larger subgraph of the KG. Specifically, our dataset has questions which require logical, quantitative, and comparative reasoning as well as their combinations. This calls for models which can: (i) parse complex natural language questions, (ii) use conversation context to resolve coreferences and ellipsis in utterances, (iii) ask for clarifications for ambiguous queries, and finally (iv) retrieve relevant subgraphs of the KG to answer such questions. However, our experiments with a combination of state of the art dialog and QA models show that they clearly do not achieve the above objectives and are inadequate for dealing with such complex real world settings. We believe that this new dataset coupled with the limitations of existing models as reported in this paper should encourage further research in Complex Sequential QA.