Interpret and Communicate

2021 ◽  
pp. 189-218
Author(s):  
Kai R. Larsen ◽  
Daniel S. Becker
Keyword(s):  
Machine Learning ◽  
Learning Process ◽  
Additional Data ◽  
Hospital Readmissions ◽  
Current Analysis ◽  
Problem Context ◽  
Time Of Admission ◽  
Time Information

Having evaluated all the measures and selected the best model for this case, and much of the machine learning process has been clarified, our understanding of the problem context is still relatively immature. That is, while we have carefully specified the problem, we still do not fully understand what drives that target. Convincing management to support the implementation of the model typically includes explaining the answers to “why,” “what,” “where,” and “when” questions embedded in the model. While the model may be the best overall possible model according to selected measures, for the particular problem related to hospital readmissions, it is still not clear why the model predicts the readmission of some patients will be readmitted and that others will not. It also remains unknown what features drive these outcomes, where the patients who were readmitted come from, or whether or not this is relevant. In this case, access to time information is also unavailable––when, so it is not relevant, but it is easy to imagine that patients admitted in the middle of the night might have worse outcomes due to tired staff or lack of access to the best physicians. If we can convince management that the current analysis is useful, we can likely also make a case for the collection of additional data. The new data might include more information on past interactions with this patient, as well as date and time information to test the hypothesis about the effect of time-of-admission and whether the specific staff caring for a patient matters.

scholarly journals Representing Deep Neural Networks Latent Space Geometries with Graphs

Algorithms ◽  
2021 ◽  
Vol 14 (2) ◽  
pp. 39
Author(s):  
Carlos Lassance ◽  
Vincent Gripon ◽  
Antonio Ortega
Keyword(s):  
Machine Learning ◽  
Neural Networks ◽  
Deep Learning ◽  
Objective Function ◽  
Learning Process ◽  
Deep Neural Networks ◽  
State Of The Art ◽  
The Core ◽  
Learning Tasks ◽  
Latent Space

Deep Learning (DL) has attracted a lot of attention for its ability to reach state-of-the-art performance in many machine learning tasks. The core principle of DL methods consists of training composite architectures in an end-to-end fashion, where inputs are associated with outputs trained to optimize an objective function. Because of their compositional nature, DL architectures naturally exhibit several intermediate representations of the inputs, which belong to so-called latent spaces. When treated individually, these intermediate representations are most of the time unconstrained during the learning process, as it is unclear which properties should be favored. However, when processing a batch of inputs concurrently, the corresponding set of intermediate representations exhibit relations (what we call a geometry) on which desired properties can be sought. In this work, we show that it is possible to introduce constraints on these latent geometries to address various problems. In more detail, we propose to represent geometries by constructing similarity graphs from the intermediate representations obtained when processing a batch of inputs. By constraining these Latent Geometry Graphs (LGGs), we address the three following problems: (i) reproducing the behavior of a teacher architecture is achieved by mimicking its geometry, (ii) designing efficient embeddings for classification is achieved by targeting specific geometries, and (iii) robustness to deviations on inputs is achieved via enforcing smooth variation of geometry between consecutive latent spaces. Using standard vision benchmarks, we demonstrate the ability of the proposed geometry-based methods in solving the considered problems.

scholarly journals Ambient Monitoring of Gait and Machine Learning Models for Dynamic and Short-Term Falls Risk Assessment in People With Dementia

2021 ◽  
Author(s):  
Navid Korhani ◽  
Babak Taati ◽  
Andrea Iaboni ◽  
Andrea Sabo ◽  
Sina Mehdizadeh ◽  
...  
Keyword(s):  
Machine Learning ◽  
Risk Assessment ◽  
Fall Risk ◽  
Learning Models ◽  
Short Term ◽  
Ambient Monitoring ◽  
Clinical Assessments ◽  
People With Dementia ◽  
Time Of Admission ◽  
Machine Learning Models

Data consists of baseline clinical assessments of gait, mobility, and fall risk at the time of admission of 54 adults with dementia. Furthermore, it includes the participants' daily medication intake in three medication categories, and frequent assessments of gait performed via a computer vision-based ambient monitoring system.

scholarly journals A low-cost machine learning process for gait measurement using biomechanical sensors

2021 ◽  
Vol 18 ◽  
pp. 100346
Author(s):  
Farah Abdel Khalek ◽  
Marc Hartley ◽  
Eric Benoit ◽  
Stephane Perrin ◽  
Luc Marechal ◽  
...  
Keyword(s):  
Machine Learning ◽  
Learning Process ◽  
Low Cost

scholarly journals The Secure Anonymised Information Linkage databank Dementia e-cohort (SAIL-DeC)

2019 ◽  
Vol 4 (3) ◽  
Author(s):  
Christian Schnier ◽  
Tim Wilkinson ◽  
Chris Orton ◽  
Laura North ◽  
Ryan Rochford ◽  
...  
Keyword(s):  
Additional Data ◽  
Diagnostic Code ◽  
Inclusion Criteria ◽  
Study Results ◽  
Dementia Research ◽  
Long Time ◽  
Study Designs ◽  
Time Information ◽  
Dementia Subtype

Introduction Dementia Platform UK (DPUK) brings together over 50 different dementia-related cohorts. Most studies have restricted follow-up times and all are based on information from people who volunteer time and data for research. Participants are therefore often not representative of the 'wider population' and generalization of results is complicated. The Secure Anonymised Information Linkage databank (SAIL) holds long-time information on every person in Wales registered with the national health service, so generalization of study results is easier; however, data management and analysis of SAIL data is not trivial. We used data from SAIL to construct an easily accessible, well described dementia e-cohort. Methods With some age restrictions, all Welsh residents for whom primary care data were available were included. Within SAIL, a table was created holding demographic information for every participant including follow-up times and several dementia indicators. Using validated diagnostic code lists, this table was linked to information on every dementia-related diagnostic event and several covariates and co-morbidites. SAIL-DeC can be modified according to varying study designs using annotated SQL-based scripts. Information on SAIL-DeC can easily be updated and linked to additional data on the SAIL database. Interactive visualisations effectively summarise cohort characteristics, aiding researchers to quickly determine cohort eligibility for dementia studies. Results From 4.4 million participants in SAIL, 1.2 million met the cohort inclusion criteria, resulting in 18.8 million person-years of follow-up. Of these, 129,650 (10%) developed all-cause dementia during follow-up, with 77,978 (60%) having dementia subtype codes. Seventy-nine percent of participants who developed dementia died during follow-up. Median survival was 12.3 years for participants diagnosed with dementia when aged 50-60, 6.8 years when aged 60-70, 4.2 years when aged 70-80 and 2.4 years when aged 80-90. Conclusions We have created a generalisable, national dementia e-cohort, aimed at facilitating epidemiological dementia research.

scholarly journals Classifying relevant video tutorials for the school’s learning management system using support vector machine algorithm

2020 ◽  
Author(s):  
Castro Mayleen Dorcas Bondoc ◽  
Tumibay Gilbert Malawit
Keyword(s):  
Machine Learning ◽  
Support Vector Machine ◽  
Learning Process ◽  
Learning Algorithm ◽  
Research Work ◽  
Supervised Machine Learning ◽  
Support Vector ◽  
Video Tutorials ◽  
Learning Management ◽  
Face To Face Instruction

Today many schools, universities and institutions recognize the necessity and importance of using Learning Management Systems (LMS) as part of their educational services. This research work has applied LMS in the teaching and learning process of Bulacan State University (BulSU) Graduate School (GS) Program that enhances the face-to-face instruction with online components. The researchers uses an LMS that provides educators a platform that can motivate and engage students to new educational environment through manage online classes. The LMS allows educators to distribute information, manage learning materials, assignments, quizzes, and communications. Aside from the basic functions of the LMS, the researchers uses Machine Learning (ML) Algorithms applying Support Vector Machine (SVM) that will classify and identify the best related videos per topic. SVM is a supervised machine learning algorithm that analyzes data for classification and regression analysis by Maity [1]. The results of this study showed that integration of video tutorials in LMS can significantly contribute knowledge and skills in the learning process of the students.

scholarly journals Cognitive Driven Multilayer Self-Paced Learning with Misclassified Samples

Complexity ◽  
2019 ◽  
Vol 2019 ◽  
pp. 1-10
Author(s):  
Qi Zhu ◽  
Ning Yuan ◽  
Donghai Guan
Keyword(s):  
Machine Learning ◽  
Learning Process ◽  
Human Learning ◽  
Learning Phase ◽  
Machine Learning Algorithms ◽  
Learning Difficulty ◽  
Learning Models ◽  
Negative Effects ◽  
Initial Learning

In recent years, self-paced learning (SPL) has attracted much attention due to its improvement to nonconvex optimization based machine learning algorithms. As a methodology introduced from human learning, SPL dynamically evaluates the learning difficulty of each sample and provides the weighted learning model against the negative effects from hard-learning samples. In this study, we proposed a cognitive driven SPL method, i.e., retrospective robust self-paced learning (R2SPL), which is inspired by the following two issues in human learning process: the misclassified samples are more impressive in upcoming learning, and the model of the follow-up learning process based on large number of samples can be used to reduce the risk of poor generalization in initial learning phase. We simultaneously estimated the degrees of learning-difficulty and misclassified in each step of SPL and proposed a framework to construct multilevel SPL for improving the robustness of the initial learning phase of SPL. The proposed method can be viewed as a multilayer model and the output of the previous layer can guide constructing robust initialization model of the next layer. The experimental results show that the R2SPL outperforms the conventional self-paced learning models in classification task.

scholarly journals An Approach to Hyperparameter Optimization for the Objective Function in Machine Learning

Electronics ◽  
2019 ◽  
Vol 8 (11) ◽  
pp. 1267 ◽  
Author(s):  
Yonghoon Kim ◽  
and Mokdong Chung
Keyword(s):  
Machine Learning ◽  
Learning Process ◽  
Learning Rate ◽  
Bayesian Optimization ◽  
Learning Performance ◽  
Batch Size ◽  
Critical Problem ◽  
Hyperparameter Optimization ◽  
Training Performance ◽  
And Performance

In machine learning, performance is of great value. However, each learning process requires much time and effort in setting each parameter. The critical problem in machine learning is determining the hyperparameters, such as the learning rate, mini-batch size, and regularization coefficient. In particular, we focus on the learning rate, which is directly related to learning efficiency and performance. Bayesian optimization using a Gaussian Process is common for this purpose. In this paper, based on Bayesian optimization, we attempt to optimize the hyperparameters automatically by utilizing a Gamma distribution, instead of a Gaussian distribution, to improve the training performance of predicting image discrimination. As a result, our proposed method proves to be more reasonable and efficient in the estimation of learning rate when training the data, and can be useful in machine learning.

A Bayesian Machine Learning Approach for Efficient Integrity Management of Steel Lazy Wave Risers

2020 ◽  
Author(s):  
Rasoul Hejazi ◽  
Andrew Grime ◽  
Mark Randolph ◽  
Mike Efthymiou
Keyword(s):  
Machine Learning ◽  
Fatigue Failure ◽  
Learning Process ◽  
Predictive Models ◽  
Structural Integrity ◽  
Degradation Mechanism ◽  
Data Driven ◽  
Practical Implementation ◽  
Integrity Management

Abstract In-service integrity management (IM) of steel lazy wave risers (SLWRs) can benefit significantly from quantitative assessment of the overall risk of system failure as it can provide an effective tool for decision making. SLWRs are prone to fatigue failure within their touchdown zone (TDZ). This failure mode needs to be evaluated rigorously in riser IM processes because fatigue is an ongoing degradation mechanism threatening the structural integrity of risers throughout their service life. However, accurately evaluating the probability of fatigue failure for riser systems within a useful time frame is challenging due to the need to run a large number of nonlinear, dynamic numerical time domain simulations. Applying the Bayesian framework for machine learning, through the use of Gaussian Processes (GP) for regression, offers an attractive solution to overcome the burden of prohibitive simulation run times. GPs are stochastic, data-driven predictive models which incorporate the underlying physics of the problem in the learning process, and facilitate rapid probabilistic assessments with limited loss in accuracy. This paper proposes an efficient framework for practical implementation of a GP to create predictive models for the estimation of fatigue responses at SLWR hotspots. Such models are able to perform stochastic response prediction within a few milliseconds, thus enabling rapid prediction of the probability of SLWR fatigue failure. A realistic North West Shelf (NWS) case study is used to demonstrate the framework, comprising a 20” SLWR connected to a representative floating facility located in 950 m water depth. A full hindcast metocean dataset with associated statistical distributions are used for the riser long-term fatigue loading conditions. Numerical simulation and sampling techniques are adopted to generate a simulation-based dataset for training the data-driven model. In addition, a recently developed dimensionality reduction technique is employed to improve efficiency and reduce complexity of the learning process. The results show that the stochastic predictive models developed by the suggested framework can predict the long-term TDZ fatigue damage of SLWRs due to vessel motions with an acceptable level of accuracy for practical purposes.

PATTERNS OF PREPOSITIONAL ATTACHMENTS — WHERE DICTIONARY SEMANTICS MEETS CORPUS STATISTICS

2000 ◽  
Vol 14 (06) ◽  
pp. 809-838 ◽  
Author(s):  
SANDA M. HARABAGIU
Keyword(s):  
Machine Learning ◽  
Information Extraction ◽  
Learning Process ◽  
Wall Street Journal ◽  
Ad Hoc ◽  
Machine Learning Techniques ◽  
Journal Articles ◽  
Wall Street ◽  
Web Documents ◽  
Learning Techniques

This paper presents a novel methodology of disambiguating prepositional phrase attachments. We create patterns of attachments by classifying a collection of prepositional relations derived from Treebank parses. As a by-product, the arguments of every prepositional relation are semantically disambiguated. Attachment decisions are generated as the result of a learning process, that builds upon some of the most popular current statistical and machine learning techniques. We have tested this methodology on (1) Wall Street Journal articles, (2) textual definitions of concepts from a dictionary and (3) an ad hoc corpus of Web documents, used for conceptual indexing and information extraction.

Sign in / Sign up

Export Citation Format

Share Document