1- Introduction

2- Related works

3- Robust coding based supervised sparse representation

4- Sparsity and robustness of our model

5- Experiment results

6- Conclusions

References

Abstract

Sparse representation based classification (SRC) has become a popular methodology in face recognition in recent years. One widely used manner is to enforce minimum -norm on coding coefficient vector, which is considered as an unsupervised sparsity constraint and usually requires high computational cost. On the other hand, supervised sparsity representation based method (SSR) realizes sparse representation classification with higher efficiency by multiple phases of representing a probe. Nevertheless, since previous SSR methods only deal with Gaussian noise, they cannot satisfy empirical face recognition application which faces wide variations. In this paper, we propose a robust supervised sparse representation (RSSR) model, which uses two-phase of robust representation to compute a sparse coding vector. Huber loss is employed as the fidelity term in the linear representation, which improves the competitiveness of correct class in the first phase. Then training samples with weak competitiveness are removed by supervised way. In the second phase, the competitiveness of correct class is further boosted by Huber loss. We compare the RSSR with other state-of-the-art methods under different conditions, including illumination variations, gesture changes, expressions, corruptions, and occlusions. Comprehensive experiments on four open databases demonstrate the robustness of RSSR and competitive performance is obtained in dealing with face images with occlusion or not.

introduction

Using biometric identification technology for verifying identity is gaining more importance, and different techniques have been developed such as Palmprint recognition[1][2] and face recognition(FR)[3][4]. FR[5][6][7]has been extensively studied for its broad application prospects in recent years, such as authentication and payment system. The primary task of FR consists of feature extraction and classification[8][9][10]. For many classifiers, feature extraction that tends to discover discriminative feature is very important, which has great influence on recognition rate. Since there is rich redundancy in a face image, low dimensional features are extracted to concisely represent the samples in training set[11][12][13][14]. So that using these features can alleviate the computational cost and improve the recognition performance of classifiers. For empirical applications of FR, various changes including lighting, expression, pose, and occlusion can be seen in a probe and which could not included in training set, which leads to the consequence that the computed feature becomes inefficient. Image segmentation can be used to help the effective representation of an image by retaining the informative parts of images[15][16].

1- Introduction

2- Methods

3- Results

4- Discussion

References

Abstract

Objective: Stuttering is a developmental disorder of speech production with a dynamic and multifactorial nature. Scientific theories mentioned the role of auditory processing disorder in stuttering. Investigating the auditory processing in stuttering would provide insights into the mechanisms of stuttering. The details of basic auditory processing in children with stuttering (CWS) continue to remain uncertain. This study aimed to investigate the auditory temporal processing (ATP) in CWS and also its relationship with the stuttering severity. Methods: The participants of this comparative cross-sectional study were 54 CWS and 63 children without stuttering (CWOS). All children were between 7 and 12 years old. ATP ability of the participants was measured using the Backward Masking (BM), Duration Pattern (DP), and Gap in Noise (GIN) tests. Then, the groups were compared in terms of ATP results. The correlation between the scores of these tests and stuttering severity was assessed. Results: According to the results, CWS showed poorer performance on DP, BM and GIN tests when compared with CWOS. Moreover, the stuttering severity had a significant negative correlation with the DP scores and percentage of correct identification scores in GIN; whereas it had a significant positive correlation with the BM thresholds. Conclusions: The results showed that some CWS have ATP disorder, which could exacerbate their stuttering. These findings highlighted the role of ATP disorder in stuttering.

Introduction

Stuttering is a common childhood developmental disorder, which affects speech fluency [1]. It may recover spontaneously, otherwise requires long-term intensive therapeutic sessions, which typically have adverse impacts on the psychological and social health of children and their family [2,3]. Stuttering is characterized by repetition of a syllable or a part of a syllable, prolongation of sounds, and frequent interruptions in speech [4]. The prevalence of stuttering in childhood is between 4% and 5%; about 1% of this population will continue to have persistent stuttering until adulthood [5]. The underlying cause of stuttering continues to remain unknown. According to current theories, stuttering is defined as a dynamic and multifactorial disorder, which occurs due to the effects of complex interactions between atypical speech motor control system, auditory processing, language and other cognitive processing, genetic, emotions, and social factors [6]. Auditory processing is a neural mechanism, which has been emphasized in numerous theories for its role in stuttering [7]. According to the Direction of Velocities into Articulators (DIVA) model, the auditory system has a role in activating and tuning speech sound maps and forming internal speech control models (feedforward and feedback models) through neural representation of sounds via processing others’ speech, receiving self-produced speech feedback, and interacting with motor cortex [8]. Proposing the “internal model and feedback-biased motor control theory,” Max et al. (2004) assumed that stuttering is due to a mismatch between predicted (feedforward) and actual (feedback) auditory consequences of speech.The researchers attributed this mismatch and speech error to the auditory feedback control pathway [9].

1- Introduction

2- Method

3- Results

4- Discussion

5- Conclusion and New Directions

References

Abstract

Purpose: This two-part (i.e., Study 1, Study 2) study investigated behavioral inhibition (BI) in preschool-age children who do (CWS) and do not (CWNS) stutter. The purpose of Study 1 was to develop the Short Behavioral Inhibition Scale (SBIS), a parent-report scale of BI. The purpose of Study 2 was to determine, based on the SBIS, differences in BI between CWS and CWNS, and associations between BI and CWS’s stuttering frequency, stuttering severity, speech-associated attitudes, and stutteringrelated consequences/reactions. Method: Participants in Study 1 were 225 CWS and 243 CWNS with the majority of them being included in Study 2. In Study 2, a speech sample was obtained for the calculation of stuttering frequency and severity, and the parents of a subset of CWS completed the Communication Attitude Test for Preschool and Kindergarten Children Who Stutter (Vanryckeghem & Brutten, 2007), and the Test of Childhood Stuttering Disfluency-Related Consequences Rating Scale (Gillam, Logan, & Pearson, 2009). Results: Study 1 analyses indicated that SBIS is a valid and reliable tool whose items assess a single, relatively homogeneous construct. In Study 2, CWS exhibited greater mean and extreme BI tendencies than CWNS. Also CWS with higher, compared to CWS with lower, BI presented with greater stuttering frequency, more severe stuttering, greater stuttering-related consequences, and more negative communication attitudes (for CWS older than 4 years of age). Conclusion: Findings were taken to suggest that BI is associated with early childhood stuttering and that the SBIS could be included as part of a comprehensive evaluation of stuttering.

Caveats

The principal component analysis indicated that SBIS items related to fear and reactivity had poor loadings and that a model without these items provided a better fit to the data. However, fear and reactivity have some degree of overlap with the construct of BI. Thus, the SBIS is less than a robust instrument for determining fear and reactivity components of BI, a limitation that should be noted by prospective users of SBIS. Also, some users might find the scoring of the test counterintuitive since higher scores indicate lower, rather than higher BI. Also, present SBIS results are based on parental responses, not unlike some findings resulting from other instruments used to test CWS, for example, the TOCS. Regardless, some clinicians as well as researchers may question the wisdom of relying on parental responses rather than, for example, more direct measurement of BI or BI-related behaviors like those of Choi et al (2013). Such questions are well taken but must await answers resulting from future research employing more direct measures. For the present, however, we would like to note Henderson and Wachs’s (2007) suggestion that although “parent report measures do contain some subjective parental components, available evidence indicates that these measures also contain a substantial objective component that does accurately assess children’s individual characteristics.” (p. 402).

Introduction

Methods

Results and discussion

Conclusion

Acknowledgements

References

Abstract

Backgrounds: An ongoing outbreak of a novel coronavirus (2019-nCoV) pneumonia hit a major city in China, Wuhan, December 2019 and subsequently reached other provinces/regions of China and other countries. We present estimates of the basic reproduction number, R0, of 2019-nCoV in the early phase of the outbreak.

Methods: Accounting for the impact of the variations in disease reporting rate, we modelled the epidemic curve of 2019-nCoV cases time series, in mainland China from January 10 to January 24, 2020, through the exponential growth. With the estimated intrinsic growth rate (g), we estimated R0 by using the serial intervals (SI) of two other well-known coronavirus diseases, MERS and SARS, as approximations for the true unknown SI.

Findings: The early outbreak data largely follows the exponential growth. We estimated that the mean R0 ranges from 2.24 (95%CI: 1.96–2.55)to 3.58 (95%CI: 2.89–4.39) associated with 8-fold to 2-fold increase in the reporting rate. We demonstrated that changes in reporting rate substantially affect estimates of R0. Conclusion: The mean estimate of R0 for the 2019-nCoV ranges from 2.24 to 3.58, and is significantly larger than 1. Our findings indicate the potential of 2019-nCoV to cause outbreaks

Introduction

The atypical pneumonia case, caused by a novel coronavirus (2019-nCoV), was first reported and confirmed in Wuhan, China on December31,2019(WorldHealthOrganization,2020a).AsofJanuary 26(17:00 GMT), 2020,therehavebeen2033 confirmedcasesof2019- nCoV infections in mainland China, including 56 deaths (National Health Commission of the People’s Republic of China, 2020). The 2019-nCoV cases were also reported in Thailand, Japan, Republic of Korea, Hong Kong, Taiwan and the US, and all of these cases were exported from Wuhan; see the World Health Organization (WHO) news release https://www.who.int/csr/don/en/ from January 14–21. The outbreak is still ongoing. A recently published preprint by Imai et al. estimated that a total of 1723 (95% CI: 427-4471) cases of 2019- nCoV infections inWuhanhad onsetof symptoms by January 12,2020 (Imai et al., 2020). The likelihood of travel related risks of disease spreading is suggested by Bogoch et al. (2020), which indicates the potentials of regional and global spread (Leung et al., 2020).

To the best of our knowledge,there is no existing peer-reviewed literature quantifying the transmissibility of 2019-nCoV as of January 22, 2020. In this study, we estimated the transmissibility of 2019-nCoV via the basic reproduction number, R0, based on the limited data in the early phase of the outbreak.

1- Introduction

2- Method

3- Results

4- Discussion

References

Assumptions to Improve Public Attitudes

The literature reporting public misinformation about stuttering and resulting stereotypes, stigma, and discrimination against those who stutter is voluminous (cf. reviews by Boyle & Blood, 2015; Gabel, 2015; Hughes, 2015; Langevin, 2015). Almost every exploratory study of public attitudes has ended with a call for education of the public about stuttering, with the stated or unstated assumption that providing accurate information about the disorder would be sufficient to improve public attitudes (e.g., Bellegarde, Mayo, St. Louis, Mayo, 2016; Cooper & Cooper, 1985; Dorsey & Guenther, 2000; Ham, 1990; Hughes, 2015; St. Louis, Przepiórka, et al., 2014; Xing Ming, Jing, Wen, & Van Borsel, 2001; Valente, St. Louis, Leahy, Hall, & Jesus, 2017). If such information were to be made available to the public, a further and often stated assumption is that the quality of life of children and adults who stutter would be improved as a result of growing up and/or living in a more accepting and informed society.

Investigations to Improve Stuttering Attitudes

Compared to studies that stopped at documenting negative attitudes, relative few investigations have attempted, experimentally, to improve them. In her review of the extant literature in 2013, Abdalla (2015) identified 15 such studies (Abdalla & St. Louis, 2014; Coleman et al., 2013; Delaney, 2001; Flynn & St. Louis, 2011; Gottwald et al., 2011; Gottwald, Kent, St. Louis, & Hartley, 2014; Hughes, Gabel, Roseman, & Daniels, 2015; Junuzović-Žunić et al., 2015; Leahy, 1994; Langevin & Prasad, 2012; Mayo, Mayo, Gentry, & Hildebrandt, 2008; McGee, Kalinowski, & Stuart, 1996; Reichel & St. Louis, 2004; Reichel & St. Louis, 2007; Snyder, 2001). Abdalla concluded that “Studies that have attempted to ameliorate negative stereotypes toward people who stutter have been inconclusive. Some have reported positive changes, while others have found either no shift in attitude or a change in the reverse direction (i.e., intervention allegedly reinforced the negative stereotypes)” (p. 117). Factors identified by Abdalla (2015) that should be considered in evaluating the research related to changing attitudes included: (a) specific constructs targeted for change; (b) prior exposure of participants to people who stutter; (c) selection criteria for participants; (d) mode of the intervention (e.g., video, coursework, or direct interaction with a stuttering person); (e) actual content of the stimuli provided; and (f) methods of design, analysis, and interpretation. Stressing that these factors are interrelated, she noted, “While the stimuli used to change attitudes may partially explain the lack of consensus in changing attitudes of fluent speakers toward stuttering, the stimuli adopted in a study must be meaningful for the target population” (p. 124, italics added). This implies that both characteristics of the interventions as well as characteristics of the persons targeted are important in explaining the success of attitude change endeavors.

1. Introduction

2. Related literature and hypotheses development

3. Data and empirical methodology

4. Empirical results

5. Robustness tests

6. Conclusion

Appendix A.

References

Abstract

High capital intensity and reliance on debt financing are among the most prominent characteristics of the shipping industry. The corporate finance literature has documented that beyond a certain threshold, leverage can hamper a firm’s ability to raise capital, and as a result, have a bearing on its corporate investment policy. The new, more restrictive, financing landscape in the shipping sector has put the management of capital structure on the spotlight as a key driver of investment policy, financial health, and thus, firm success. In this paper, we examine for the first time the link between the financing policy of shipping companies and their corporate investment decisions. We focus on the impact of deviations from target capital structure on mergers and acquisitions (M&A); an increasingly important corporate growth vehicle for shipping companies, with directly measurable outcomes. Deviations from optimal leverage display a strong association with the likelihood to consummate acquisitions, deal size, the financing method as well as the M&A outcome. Higher debt levels are shown to have a negative effect on acquisitiveness and a positive effect on the quality of corporate investment; a pattern with direct policy implications for shipping companies, their management teams, and shareholders.

Introduction

The shipping industry is one of the most capital-intensive. Shipping companies’ CAPEX-to-Assets ratio averaged around 8% between 1990 and 2018, placing the sector among the top 8th percentile of all sectors.2 In deep-sea freight transportation, capital investments are almost exclusively associated with vessel purchases, which are in turn financed with debt at more than 40% of capital employed for the average firm (Drobetz et al., 2013). The idiosyncratic characteristics of the shipping industry including the high asset tangibility and equity risk environment, have naturally led to more debt-driven capital structures. More recently, the sector has reached critical levels of borrowing with the world’s top 40 banks having a $345bil exposure to the shipping industry (Petrofin, 2018), and an estimate of $150bil of loans provided by European banks alone considered distressed (Reuters, 2017).3 The high capital ntensity and reliance on debt financing associated with shipping companies, coupled with the financial constraints brought forward by the new financing environment, suggest that the success of a shipping company is highly sensitive to its debt policy since deviations from target capital structures can lead to a high cost of financial distress (Drobetz et al., 2013). The corporate finance literature has provided ample evidence that a high degree of leverage in a firm’s capital structure can hamper its ability to raise more capital (Hovakimian et al., 2001; Fama and French, 2002; Flannery and Rangan, 2006) and, as a result, limit its flexibility in devising corporate investment policy (Harford et al., 2009; Uysal, 2011; DeAngelo et al., 2011). The impact of financial leverage on corporate investment policy can be more pronounced in the shipping industry due to its idiosyncrasies, making it a natural laboratory to study the effects financing policy on corporate investment.

1. Introduction

2. The future shape of human-artificial intelligence nexus and its environmental costs

3. The environmental costs of machine learning research

4. The environmental costs of gratuitous generalisation capabilities

5. Conclusion

Acknowledgment

References

Abstract

The environmental costs and energy constraints have become emerging issues for the future development of Machine Learning (ML) and Artificial Intelligence (AI). So far, the discussion on environmental impacts of ML/AI lacks a perspective reaching beyond quantitative measurements of the energy-related research costs. Building on the foundations laid down by Schwartz et al. (2019) in the GreenAI initiative, our argument considers two interlinked phenomena, the gratuitous generalisation capability and the future where ML/AI performs the majority of quantifiable inductive inferences. The gratuitous generalisation capability refers to a discrepancy between the cognitive demands of a task to be accomplished and the performance (accuracy) of a used ML/AI model. If the latter exceeds the former because the model was optimised to achieve the best possible accuracy, it becomes inefficient and its operation harmful to the environment. The future dominated by the non-anthropic induction describes a use of ML/AI so all-pervasive that most of the inductive inferences become furnished by ML/AI generalisations. The paper argues that the present debate deserves an expansion connecting the environmental costs of research and ineffective ML/AI uses (the issue of gratuitous generalisation capability) with the (near) future marked by the all-pervasive Human-Artificial Intelligence Nexus.

Introduction

Conceived as a scholarly discipline, ML seeks to develop ‘tools-for-optimal-action’. Given a task and evidence that can facilitate its mastering, the ‘tool-for-optimal-action’ earns its rank by being able to generalise. Such a tool then supports inferences which can generalise beyond the evidence (training data), i.e. can run inferences on new samples, provided that these come from the same, or a sufficiently similar, probability distribution as the evidence (training data). The discipline has a twofold epistemic aim. First, its theoretical purview, established by statistical learning theory, involves formal assumptions about the learning that leads to generalisations (Kawaguchi, Kaelbling, & Bengio, 2019; Vapnik, 1995). Second, from the empirical viewpoint, the discipline seeks to improve the accuracy of inferences that furnish the acquired generalisations. At the moment, the cross-fertilisation between the two sub goals seems to be rather recalcitrant creating the following asymmetry. Although investing heavily in the theoretical research (e.g. cf. Arjovsky, Bottou, Gulrajani, & Lopez-Paz, 2019; Bartlett, Foster, & Telgarsky, 2017; Kawaguchi et al., 2019; Neyshabur, Bhojanapalli, McAllester, & Srebro, 2017; Zhang, Bengio, Hardt, Recht, & Vinyals, 2017), the field remains dominated by the second sub goal. As strong empirical results outpaced mature theoretical understanding, task-specific, accuracy-tracking leaderboards became the go-to measure for assessing the field’s epistemic progress.

The emphasis put on a single objective – accuracy – inspires a naïve idea that leaderboards are like ladders. The higher the rung, the closer we are to alleviating our cognitive burden by employing almost perfect ‘tools-for-optimal-action’ to carry out all sorts of tasks. If we construe the epistemic aim of ML as understanding generalisations, incentivising leaderboards produces troubles for the discipline itself and, quite strikingly, for the environment as well. The issue concerns the cost of computational resources that enable climbing to ever higher positions on the leaderboards. In scenarios where theoretical understanding lags behind empirical results, a new state-of-the-art (SOTA) usually emerges from trial and error experimentations often producing quite arbitrary heuristics. Faced with the theoretical lacuna, practitioners confront the temptation of post-hoc speculations that might assume the role usually played by theoretical explanations (cf. Lipton & Steinhardt, 2019). When occurring alongside (accidental) misattributions of the sources of empirical gains, f.e. reporting improvements from neural architecture changes when, in reality, they stem from hyperparameter tuning (ibid.), the following might ensue. Instead of achieving the epistemic aim of understanding generalisations, leaderboards might merely encourage post-hoc hypotheses fitted to the results of quite arbitrary heuristics. The incredibly fast pace of the leaderboards climbs, natural language processing (NLP) is among the best of present examples (cf. Strubell, Ganesh, & McCallum, 2019 for an estimation of the SOTA NLP’s environmental costs), makes such bad practices a siren song, which could considerably hamper the discipline’s twofold epistemic goal.

Introduction

Materials and Methods

Results

Discussion

Conclusions

References

Abstract

Background: Stroke mimic is a medical condition presenting with acute neurological deficit and simulate real stroke. The objective of this study was to evaluate the frequency and the various etiologies of stroke mimics in our center. Methods: We retrospectively reviewed the Thrombolysis Alert registry and we studied the frequency and characteristics of patients with stroke mimic. Results: Among 673 patients who were admitted to the emergency department within 4.5 hours for sudden focal neurological deficit suggestive of acute stroke, 105 patients (15.6 %) had a stroke mimic. The mean age of patients with mimics and brain strokes were 66.3 and 64.8, respectively. The mean Onset-to-door time was 136.82 minutes and the mean door-to-imaging time was 32.63 minutes in stroke mimics. Seizure (28.5%) was the most common diagnosis of stroke mimics followed by conversion disorder (25.7%). Conclusions: Stroke mimic is frequent and heterogeneous entity that can be difficult to identify. Fortunately, most previous studies show no harmful effects when using thrombolysis in a stroke mimic.

Introduction

Brain Stroke is a major cause of mortality and morbidity. In Morocco, the prevalence rate of stroke in all agesis estimated at 292 per 100,000. The management of ischemic stroke has seen major advances with intravenous thrombolysis (IVT) and endovascular thrombectomy (EVT). IVT has been the standard treatment for acute ischemic stroke (AIS) patients within 4.5 hours of symptom onset. Thrombolysis with tissue plasminogen activator is the first effective approved therapy for potentially disabling AIS. Recently there is more evidence suggesting that other thrombolytics such as tenecteplase may be even more effective. EVT is another effective therapy; it reduces the risk of death or dependency in patients with large vessel occlusion if applied within 6 hours. The recently published DAWN and DEFUSE 3 trials have shown that EVT is also beneficial 6-24 hours after the onset of symptoms in selected patients with favorable imaging brain. The diagnosis of acute ischemic stroke is usually easy by carrying out physical evaluation and studying images of the brain, however several disorders, manifested by an acute neurological deficit can mimic AIS. Stroke mimic term is applied in a clinical evaluation, describing those nonvascular conditions that simulate stroke, namely those presenting with sudden neurological deficit that resembles a stroke but is not related to a vascular cause. Stroke mimics are a major diagnostic challenge. The aim of this study is to evaluate the relative frequency and the various etiologies of stroke mimics in patients with clinical signs of stroke admitted in our emergency room for “Thrombolysis Alert.”

1- Introduction

2- Previous work

3- Data transcription and annotation

4- Detecting stuttering events

5- Features of the classifiers used to detect stuttering events

6- Automatic speech recognition system

7- Experiments

8- Conclusion

References

Abstract

Stuttering is a speech disorder that, if treated during childhood, may be prevented from persisting into adolescence. A clinician must first determine the severity of stuttering, assessing a child during a conversational or reading task, recording each instance of disfluency, either in real time, or after transcribing the recorded session and analysing the transcript. The current study evaluates the ability of two machine learning approaches, namely conditional random fields (CRF) and bi-directional long-short-term memory (BLSTM), to detect stuttering events in transcriptions of stuttering speech. The two approaches are compared for their performance both on ideal hand-transcribed data and also on the output of automatic speech recognition (ASR). We also study the effect of data augmentation to improve performance. A corpus of 35 speakers’ read speech (13K words) was supplemented with a corpus of 63 speakers’ spontaneous speech (11K words) and an artificially-generated corpus (50K words). Experimental results show that, without feature engineering, BLSTM classifiers outperform CRF classifiers by 33.6%. However, adding features to support the CRF classifier yields performance improvements of 45% and 18% over the CRF baseline and BLSTM results, respectively. Moreover, adding more data to train the CRF and BLSTM classifiers consistently improves the results.

Introduction

Stuttering, also known as stammering, is a speech communication disorder that can have severe social, educational and emotional maladjustment consequences, not only for the people who stutter but also for their families [1, 2]. It is presumed that early intervention is best to offset potential later impacts of having a stutter on one’s psycho-social and communication developments [3]. During the assessment phase, clinicians carefully measure the stuttering events to determine if the stuttering is normal disfluency, borderline stuttering or beginning stuttering [4]. There are several approaches to determine stuttering severity. The fluency of very young children is commonly assessed through a conversational task, whereas for children older than seven years, a reading task may be used [5, 6]. The clinician asks the child to read from a passage, and then records each instance of disfluency while the child is reading. Clearly, this process is extremely dependent on the clinician’s experience [7, 8, 9, 10].

Acknowledgements

References

Abstract

A novel bat-origin coronavirus emerged in Wuhan, China in December 2019 and continues to spread across China and the world. At the time of writing, a massive global response has been implemented to control the disease as it spreads from person to person. Yet the high-risk human-wildlife interactions and interfaces that led to the emergence of SARSCoV and of 2019-nCoV continue to exist in emerging disease hotspots globally. To prevent the next epidemic and pandemic related to these interfaces, we call for research and investment in three areas: 1) surveillance among wildlife to identify the high-risk pathogens they carry; 2) surveillance among people who have contact with wildlife to identify early spillover events; and 3) improvement of market biosecurity regarding the wildlife trade. As the emergence of a novel virus anywhere can impact the furthest reaches of our connected world, international collaboration among scientists is essential to address these risks and prevent the next pandemic.

The emergence of a novel coronavirus, 2019-nCoV in Wuhan December 2019 has led to a global response to the first epidemic of this decade. It has also highlighted two key issues: First, China’s rapid and efficient capacity to identify and investigate a newly emerging disease; and second, our continued global vulnerability to epidemics and pandemics. From the date of the first cluster of cases admitted to a local hospital on December 27, 2019, Chinese scientists identified this disease as a new syndrome, discovered the pathogen as the cause, and reported its genetic sequence to the world in less than 14 days [1,2]. At the time of writing, we have detailed information on its relationship to other bat coronaviruses, many of which were discovered in a collaboration among EcoHealth Alliance, the Wuhan Institute of Virology, and Duke NUS [3]. We also have data from experimental cell line infections [4], on the clinical findings [5], the epidemiology of viral transmission [6], and on its spread to other countries. These rapid and successful results are a resounding endorsement of China’s emergence as a 21st Century scientific superpower, and of a country with a modern and efficient public health system, technically supported by the China CDC with a system of Provincial and City CDCs that are trained in effective outbreak investigation and control [7].

1. Can AI contribute to sales performance?

2. Foundations of AI

3. The AI-enabled B2B sales funnel

4. Managerial considerations for maximizing collaborative intelligence

5. AI will enhance, not replace

References

Abstract

The B2B sales process is undergoing substantial transformations fueled by advances in information and communications technology, specifically in artificial intelligence (AI). The premise of AI is to turn vast amounts of data into information for superior knowledge creation and knowledge management in B2B sales. In doing so, AI can significantly alter the traditional human-centric sales process. In this article, we describe how AI affects the B2B sales funnel. For each stage of the funnel, we describe key sales tasks, explain the specific contributions AI can bring, and clarify the role humans play. We also outline managerial considerations to maximize the contributions from AI and people in the context of B2B sales.

Can AI contribute to sales performance?

B2B sales managers have always been concerned with serving their clients at every stage of thesales process. However, much has changed since the days of stereotypical salesperson Willy Loman, the protagonist in Arthur Miller’s 1949 play Death of a Salesman. As the opening quote reveals, Loman’s travels took him to many cities in which he could rely on personal, emotional, and professional connections with his clients. In the process, Loman and his fellow salespeople mainly relied on manual, analog technologies (e.g., maps, landline telephones) to enable their B2B sales.

The launch of the world’s first mobile telephone certainly seemsdand wasda long time ago. Many technological innovations have shaped B2B sales since, especially profound advances in information and communications technology (Morkunas, Paschen, & Boon, 2019; Paschen, Pitt, & Kietzmann, 2019), digitization, and artificial intelligence (AI) (Gupta, Keen, Shah, Verdier, & Walutowy, 2017; Syam & Sharma, 2018). In particular, AI systemsddefined as information systems that act intelligently (Russell, 1997; Russell & Norvig, 2016)dare substantially different than technological advancements of the past (Kietzmann, Paschen, & Treen, 2018). While previous technological innovations, including the advent of the telephone, computers, or the internet, helped with collecting, aggregating, processing, or communicating data (Paschen, Pitt, & Kietzmann, 2020), decision making always rested with humans. Today, emerging technologies such as AI can significantly alter the interactions between humans and machines (Marr, 2016), enabling computers to solve problems with minimal or no human intervention (Syam & Sharma, 2018).

Introduction

Methods

Results

Discussion

Data sharing

Supplementary Material

References

Summary

Background Since Dec 31, 2019, the Chinese city of Wuhan has reported an outbreak of atypical pneumonia caused by the 2019 novel coronavirus (2019-nCoV). Cases have been exported to other Chinese cities, as well as internationally, threatening to trigger a global outbreak. Here, we provide an estimate of the size of the epidemic in Wuhan on the basis of the number of cases exported from Wuhan to cities outside mainland China and forecast the extent of the domestic and global public health risks of epidemics, accounting for social and non-pharmaceutical prevention interventions.

Methods We used data from Dec 31, 2019, to Jan 28, 2020, on the number of cases exported from Wuhan internationally (known days of symptom onset from Dec 25, 2019, to Jan 19, 2020) to infer the number of infections in Wuhan from Dec 1, 2019, to Jan 25, 2020. Cases exported domestically were then estimated. We forecasted the national and global spread of 2019-nCoV, accounting for the effect of the metropolitan-wide quarantine of Wuhan and surrounding cities, which began Jan 23–24, 2020. We used data on monthly flight bookings from the Official Aviation Guide and data on human mobility across more than 300 prefecture-level cities in mainland China from the Tencent database. Data on confirmed cases were obtained from the reports published by the Chinese Center for Disease Control and Prevention. Serial interval estimates were based on previous studies of severe acute respiratory syndrome coronavirus (SARS-CoV). A susceptible-exposed-infectious-recovered metapopulation model was used to simulate the epidemics across all major cities in China. The basic reproductive number was estimated using Markov Chain Monte Carlo methods and presented using the resulting posterior mean and 95% credibile interval (CrI).

Findings In our baseline scenario, we estimated that the basic reproductive number for 2019-nCoV was 2·68 (95% CrI 2·47–2·86) and that 75 815 individuals (95% CrI 37 304–130 330) have been infected in Wuhan as of Jan 25, 2020. The epidemic doubling time was 6·4 days (95% CrI 5·8–7·1). We estimated that in the baseline scenario, Chongqing, Beijing, Shanghai, Guangzhou, and Shenzhen had imported 461 (95% CrI 227–805), 113 (57–193), 98 (49–168), 111 (56–191), and 80 (40–139) infections from Wuhan, respectively. If the transmissibility of 2019-nCoV were similar everywhere domestically and over time, we inferred that epidemics are already growing exponentially in multiple major cities of China with a lag time behind the Wuhan outbreak of about 1–2 weeks.

Interpretation Given that 2019-nCoV is no longer contained within Wuhan, other major Chinese cities are probably sustaining localised outbreaks. Large cities overseas with close transport links to China could also become outbreak epicentres, unless substantial public health interventions at both the population and personal levels are implemented immediately. Independent self-sustaining outbreaks in major cities globally could become inevitable because of substantial exportation of presymptomatic cases and in the absence of large-scale public health interventions. Preparedness plans and mitigation interventions should be readied for quick deployment globally

1. Introduction and research motivation

2. Research on Doom using the VizDoom Game-AI research platform

3. Direct future prediction (DFP)

4. Direct future prediction (DFP) and asynchronous advantage Actor-Critic (A3C)

5. Advantage Actor-critic (A2C) and advantage Actor-Critic- long Short-Term memory (A2C-LSTM)

6. Asynchronous advantage Actor-Critic (A3C) and deep recurrent Q-Network (DRQN)

7. Direct future prediction (DFP) and deep recurrent Q-Network (DRQN)

8. Deep Q-Network (DQN) and deep recurrent Q-Network (DRQN)

9. Dueling deep Q-Network (DDQN) and double deep Q-Network (DDQN)

10. C51_DDQN and Reinforce

11. Techniques with better performance

12. Discussion and future work

Declaration of Competing Interest

Acknowledgment

References

Abstract

Artificial Intelligence in the form of machine learning is employed in games to control non-human computer-players, agents or bots. However, most of these games such as Atari took place in 2D environments that were fully observable to the agents. Currently, it is of extreme significance to employ such machine learning techniques and methods in 3D environments such as Doom. Therefore, In this paper, we train agents on the health gathering scenario of the classical first-person shooter game Doom by first presenting the Direct Future Prediction to train an agent that uses a simple architecture with no additional supervisory signals, then differentiate and compare the performance of the agents trained by using several different machine learning techniques, and the AI reinforcement learning platform ‘VizDoom’, a 3D partially observable environment, with interesting enhanced properties that makes agents to stand out from inbuilt AI agents and human players. We have continued to use computer games as a benchmark for the performance of AI as having been so successful in the past. We also compared the results of our findings to conclude the performance of the agents trained with different machine learning techniques. The agents performed well against both human players and inbuilt game agents.

Introduction and research motivation

In the last few decades, due to the progress in artificial intelligence, a revolution and sudden change have been observed in the technology both in hardware and software [1]. This change is seeping and taking over in our lives up to a certain extent, affecting how we live, work and entertain ourselves such as employing domestic robots servants, healthcare uses, electronic trading, remote sensing, expert systems, traffic control systems, autonomously-powered self-driving vehicles, and from behavioral algorithms to suggestive searches [2], etc. In the same way, gaming is a widely recognized part of our cultural landscape and as old as our human ancestors. The earliest computers were very slow and the interaction with the user was limited to basic principles. In the early ’40s, computers evolved, and programmers started to develop new virtual worlds and surprising ways of interaction between the user and the machine [3]. But now due to advancements in technology such as GPU’s[4], TPU’s[5] and the revolution in deep neural networks [6] it has become possible for artificial intelligence to stepin in video games as well where massive graphical data in the form of frames, or to be more specific a huge amount of multidimensional data is required to be processed and execute [7]. In the recent past, machine learning techniques and methods were employed in Atari games for training agents, where later, the agents performed on 49 different Atari games with better and improved results. However, most of these Atari games took place in 2D environments that were fully observable to the agents [8]. Currently, it is of extreme significance to employ such machine learning techniques and methods in 3D environments such as Doom [9] a first-person-shooter game shown in fig. 1, Starcraft [10] a third person shooter game based on real-time strategies, and sandbox open-world games such as Grand Theft Auto V and Minecraft [11] because the research community in AI think and consider that computer video games are the best test-beds for testing different artificial intelligence techniques, methods, and algorithms before evaluating them in real-world life. Thus, in this paper, state-of-the-art machine learning techniques that were before partially tested in 2D environments are now employed in a 3D environment known as Doom, to train, differentiate and compare agents performances, such as advantage actor-critic (A2C) [12], advantage actor-critic long short-term memory (A2C-LSTM) [13], asynchronous advantage actor-critic (A3C) [14], Deep Q-network (DQN) [15], Deep recurrent Q-network (DRQN) [16], Double deep Q-network (DDQN) [17], C51-DDQN [18], Dueling deep Q-network (DDQN) [19], and Reinforce [20] whereafter applying most of the agents are found useful and effective. In addition, this paper presents one of the 4 best techniques that performed well on the VizDoom AI platform [21]. It was suggested that making such research available is beneficial for the community researching on first-person-shooter games which may set up a base for further research and improvement.