| 1. |
- Alfalahi, Alyaa, et al.
(författare)
-
Expanding a dictionary of marker words for uncertainty and negation using distributional semantics
- 2015
-
Ingår i: EMNLP 2015 - 6th International Workshop on Health Text Mining and Information Analysis, LOUHI 2015 : Proceedings of the Workshop - Proceedings of the Workshop. - : Association for Computational Linguistics. - 9781941643327 ; , s. 90-96
-
Konferensbidrag (refereegranskat)abstract
- Approaches to determining the factuality of diagnoses and findings in clinical text tend to rely on dictionaries of marker words for uncertainty and negation. Here, a method for semi-automatically expanding a dictionary of marker words using distributional semantics is presented and evaluated. It is shown that ranking candidates for inclusion according to their proximity to cluster centroids of semantically similar seed words is more successful than ranking them according to proximity to each individual seed word.
|
|
| 2. |
- di Lucia, Lorenzo, 1977, et al.
(författare)
-
Decision-making fitness of methods to understand Sustainable Development Goal interactions
- 2022
-
Ingår i: Nature Sustainability. - : Springer Science and Business Media LLC. - 2398-9629. ; 5:2, s. 131-138
-
Tidskriftsartikel (refereegranskat)abstract
- The integrated nature of the Sustainable Development Goals (SDGs) presents a challenge to implementing the 2030 Agenda. Analytical methods to support decision-makers are often developed without explicitly incorporating decision-makers’ views and experience. Here, we investigate whether existing methods are fit-for-purpose in supporting decision-makers at national and subnational levels. We identify prominent methods for SDG interaction analysis, which we then evaluate by engaging directly (via a survey and interviews) with method developers and decision-makers in Sweden. We find that decision-makers prioritize methods that are simple and flexible to apply and able to provide directly actionable and understandable results. They are less concerned with the accuracy, precision, completeness or quantitative nature of the knowledge. Prominent categories of methods include self-assessment, expert judgement, literature-based, statistical analyses and modelling. Interviewed decision-makers consider these methods in line with the features prioritized in the survey but highlight low performance on features they value highly, such as the extent to which results are actionable and overall ease of use. Methods developers have limited awareness of decision-makers’ priorities and requirements, so hindering methodological advancement. They should focus on the practical value of applications to support decision-makers, resource-constrained organizations and those seeking to evaluate multiple cases.
|
|
| 3. |
- Memedi, Mevludin, et al.
(författare)
-
Validity and responsiveness of at-home touch-screen assessments in advanced Parkinson's disease
- 2015
-
Ingår i: IEEE journal of biomedical and health informatics. - : Institute of Electrical and Electronics Engineers (IEEE). - 2168-2194 .- 2168-2208. ; 19:6, s. 1829-1834
-
Tidskriftsartikel (refereegranskat)abstract
- The aim of this study was to investigate if a telemetry test battery can be used to measure effects of Parkinson’s disease (PD) treatment intervention and disease progression in patients with fluctuations. Sixty-five patients diagnosed with advanced PD were recruited in an open longitudinal 36-month study; 35 treated with levodopa-carbidopa intestinal gel (LCIG) and 30 were candidates for switching from oral PD treatment to LCIG. They utilized a test battery, consisting of self-assessments of symptoms and fine motor tests (tapping and spiral drawings), four times per day in their homes during week-long test periods. The repeated measurements were summarized into an overall test score (OTS) to represent the global condition of the patient during a test period. Clinical assessments included ratings on Unified PD Rating Scale (UPDRS) and 39-item PD Questionnaire (PDQ-39) scales. In LCIG-naïve patients, mean OTS compared to baseline was significantly improved from the first test period on LCIG treatment until month 24. In LCIG-non-naïve patients, there were no significant changes in mean OTS until month 36. The OTS correlated adequately with total UPDRS (rho = 0.59) and total PDQ-39 (0.59). Responsiveness measured as effect size was 0.696 and 0.536 for OTS and UPDRS respectively. The trends of the test scores were similar to the trends of clinical rating scores but dropout rate was high. Correlations between OTS and clinical rating scales were adequate indicating that the test battery contains important elements of the information of well-established scales. The responsiveness and reproducibility were better for OTS than for total UPDRS.
|
|
| 4. |
- Andersson, Bo, Docent, 1965-, et al.
(författare)
-
Developing M-Services : Lessons Learned From the Developers Perspective.
- 2007
-
Ingår i: Communications of the Association for Information Systems. - : Association for Information Systems. - 1529-3181 .- 1529-3181. ; 20:38, s. 605-620
-
Tidskriftsartikel (refereegranskat)abstract
- In recent years the Swedish Police Force (SPF) have encountered greater demands on availability and 24/7 services when dealing with errands that are regarded as low priority compared to regular police work, e.g. collecting tips from the public. One attempt to meet these increasing demands was the development of a mobile communications platform that allowed the public to communicate easily with the SPF using their own mobile phones by sending SMS and MMS. The focus of this paper is on the early phases of development of this m-service, in particular, on the specific technical issues such as interoperability and standards used by the actors on the scene affecting the development of mobile information systems. The learning experiences are as follows: First, mobile communication platforms have a large potential for contributing to the field of emergency management information systems since they can be based on open and nationally accepted standards. Second, global and national standards for sending multimedia messages are not always truly standardized. Operators and mobile phone manufacturers make minor alterations and interpretations of the standard and thereby some of the benefits found in standards disappear. Third, when developing mobile information systems we suggest and recommend that the analysis phase should be enhanced compared to traditional system development, and it should address the interoperability between mobile phones on one hand and operators on the other hand.
|
|
| 5. |
|
|
| 6. |
- Hrastinski, Stefan
(författare)
-
Dimensions of synchronous online education
- 2007
-
Ingår i: Principles of Effective Online Teaching. - Santa Rosa, California : Informing Science Press. - 193288601X ; , s. 105-119, s. 105-119
-
Bokkapitel (populärvet., debatt m.m.)
|
|
| 7. |
- Sadok, Moufida, et al.
(författare)
-
Understanding Security Practices Deficiencies: A Contextual Analysis
- 2015
-
Ingår i: Human Aspects of Information Security and Assurance Conference Proceedings. - 9781841023885 ; , s. 151-160
-
Konferensbidrag (refereegranskat)abstract
- This paper seeks to provide an overview of how companies assess and manage security risks in practice. For this purpose we referred to data of security surveys to examine the scope of risk analysis and to identify involved entities in this process. Our analysis shows a continuous focus on data system security rather than on real world organizational context as well as a prevalent involvement of top management and security staff in risk analysis process and in security policy definition and implementation. We therefore suggest that three issues need to be further investigated in the field of information security risk management in order to bridge the gap between design and implementation of secure and usable systems. First, there is a need to broaden the horizon to consider information system as human activity system which is different from a data processing system. Second, the involvement of relevant stakeholders in context for risk analysis leads to better appreciation of security risks. Third, it is necessary to develop ad-hoc tools and techniques to facilitate discussions and dialogue between stakeholders in risk analysis context.
|
|
| 8. |
- Meding, Isak, et al.
(författare)
-
You can have your ensemble and run it too - Deep Ensembles Spread Over Time
- 2023
-
Ingår i: Proceedings - 2023 IEEE/CVF International Conference on Computer Vision Workshops, ICCVW 2023. - : IEEE COMPUTER SOC. ; , s. 4022-4031
-
Konferensbidrag (refereegranskat)abstract
- Ensembles of independently trained deep neural networks yield uncertainty estimates that rival Bayesian networks in performance. They also offer sizable improvements in terms of predictive performance over single models. However, deep ensembles are not commonly used in environments with limited computational budget - such as autonomous driving - since the complexity grows linearly with the number of ensemble members. An important observation that can be made for robotics applications, such as autonomous driving, is that data is typically sequential. For instance, when an object is to be recognized, an autonomous vehicle typically observes a sequence of images, rather than a single image. This raises the question, could the deep ensemble be spread over time?In this work, we propose and analyze Deep Ensembles Spread Over Time (DESOT). The idea is to apply only a single ensemble member to each data point in the sequence, and fuse the predictions over a sequence of data points. We implement and experiment with DESOT for traffic sign classification, where sequences of tracked image patches are to be classified. We find that DESOT obtains the benefits of deep ensembles, in terms of predictive and uncertainty estimation performance, while avoiding the added computational cost. Moreover, DESOT is simple to implement and does not require sequences during training. Finally, we find that DESOT, like deep ensembles, outperform single models for out-of-distribution detection.
|
|
| 9. |
- Hrastinski, Stefan, et al.
(författare)
-
Design exemplars for synchronous e-learning : A design theory approach
- 2010
-
Ingår i: Computers and education. - Amsterdam : Elsevier BV. - 0360-1315 .- 1873-782X. ; 55:2, s. 652-662
-
Tidskriftsartikel (refereegranskat)abstract
- Synchronous e-learning has received much less research attention, as compared with asynchronous e-learning. Practitioners that consider using and designing synchronous e-learning are in urgent need of guidance. In order to address this need, we propose design exemplars for synchronous e-learning. They are directed towards a primary constituent community of teachers, administrators, managers and developers of e-learning. The exemplars have been theoretically as well as empirically grounded through cross-case analyses of studies conducted between 2003 and 2006. Moreover, the exemplars have been evaluated by conducting focus group sessions with experienced practitioners having experience of using and developing e-learning. Strong support was identified for each design exemplar. The exemplars can be used as research hypotheses and be tested in future design research. (C) 2010 Elsevier Ltd. All rights reserved.
|
|
| 10. |
- Balkenius, Christian, et al.
(författare)
-
Elements of cognition for general intelligence
- 2023
-
Ingår i: Artificial General Intelligence : 16th International Conference, AGI 2023, Stockholm, Sweden, June 16–19, 2023, Proceedings - 16th International Conference, AGI 2023, Stockholm, Sweden, June 16–19, 2023, Proceedings. - 1611-3349 .- 2945-9133 .- 0302-9743 .- 2945-9141. - 9783031334689 - 9783031334696 ; 13921, s. 11-20
-
Konferensbidrag (refereegranskat)abstract
- What can artificial intelligence learn from the cognitive sciences? We review some fundamental aspects of how human cognition works and relate it to different brain structures and their function. A central theme is that cognition is very different from how it is envisioned in classical artificial intelligence which offers a novel path toward intelligent systems that in many ways is both simpler and more attainable. We also argue that artificial intelligent systems takes more than a single silver bullet. It requires a large number of interacting subsystem that are coupled to both the body and to the environment. We argue for an approach to artificial general intelligence based on a faithful reproduction of known brain processes in a system-level model that incorporates a large number of components modelled after the human brain.
|
|
