Phone sensors detect gait changes during drinking episodes.
problem Detecting gait changes in heavy drinkers during natural drinking occasions.
method Used phone sensors to collect gait data, computed features, and trained an artificial neural network model.
result Phone sensor features correlate highly with estimated blood alcohol concentration (eBAC).
Paper proposes real-time health monitoring using wearable sensors.
problem Real-time health monitoring using wearable sensors.
method Learning shallow detection cascades for real-time wearable-phone or wearable-phone-cloud systems.
result Cigarette smoking detection from actigraphy and respiration data.
Understanding the spatiotemporal distribution of people within a city is crucial to many planning applications. Obtaining data to create required knowledge, currently involves costly survey methods. At the same time ubiquitous mobile sensors from personal GPS devices to mobile phones are collecting massive amounts of d…
Framework transfers inertial motion across domains without labeled data.
problem Inertial measurements are sensitive to sensor placement and motion dynamics.
method Extracts domain-invariant features and transfers them to new domains.
result Framework converts raw IMU sequences into accurate inertial trajectories.
System detects social interactions in crowds using mobile phone sensors.
problem Detecting social interactions in crowded settings.
method Multi-modal mobile sensing (BLE, accelerometer, gyroscope) and machine learning.
result 77.8% precision and 86.5% recall for predicting social interactions.
Research has proven that stress reduces quality of life and causes many diseases. For this reason, several researchers devised stress detection systems based on physiological parameters. However, these systems require that obtrusive sensors are continuously carried by the user. In our paper, we propose an alternative a…
HAR-Net combines deep features with traditional hand-crafted features for better human activity recognition.
problem Challenges in traditional HAR methods, especially feature extraction.
method Combines deep learning and traditional feature engineering.
result Performance improvement of 0.9% compared to traditional SVM.
Study provides guidelines for smartphone-based transportation mode detection.
problem Developing a reliable transportation mode detection system using smartphone sensors.
method Detailed dataset construction, sensor relevance analysis, and unknown user detection.
result Demonstrated the feasibility and effectiveness of smartphone-based TMD.
We propose a sparse-coding framework for activity recognition in ubiquitous and mobile computing that alleviates two fundamental problems of current supervised learning approaches. (i) It automatically derives a compact, sparse and meaningful feature representation of sensor data that does not rely on prior expert know…
SilentPhone identifies opportune times to silence phones to reduce interruptions.
problem Inappropriate phone notifications cause interruptions for users and others.
method Data-driven approach using past phone log data to infer unavailability.
result Identifies opportune moments for call interruptions and generates silent mode rules.
Tensor factorization uncovers hidden patterns in student behavior data.
problem Discovering low-dimensional structure in high-dimensional behavioral data.
method Non-negative tensor factorization applied to wearable sensor data.
result Tensor factorization reveals clusters of students with different behaviors.
Bluetooth data predicts depression severity, showing 18.8% extra variance.
problem Predicting depressive symptom severity using Bluetooth data.
method Extracted 49 Bluetooth features from NBDC data, used linear mixed-effect and hierarchical Bayesian linear regression models.
result Hierarchical Bayesian model achieved best prediction metrics (R2=0.526, RMSE=3.891).
Paper accelerates nonlinear mapping in online systems with lower time complexity.
problem Speeding up nonlinear mapping in online systems.
method Integrates an acceleration module into Dendrite Net (DD) to reduce time complexity.
result DD with AC has lower time complexity while maintaining nonlinear mapping and system identification properties.
The study detects and classifies touch gestures with high accuracy.
problem Detecting and classifying touch gestures from touch screens.
method Supervised learning techniques using a capacitive sensor array to record touch and swipe gestures.
result Logistic Regression models achieved over 95% accuracy for all gesture types.
Paper presents a robust model to improve prediction accuracy for real-life mobile phone data.
problem Noisy instances in real-life mobile phone data affect model accuracy.
method Identify and eliminate noisy instances using naive Bayes and Laplace estimators, then build a decision tree model.
result The robust model improves prediction accuracy as shown by experimental results.
The paper calibrates phone likelihoods in speech recognition systems.
problem Improving the calibration of phone likelihoods in speech recognition systems.
method Reduced Kaldi's DNN shared pdf-id posteriors to phone likelihoods, evaluated using a calibration sensitive metric, and improved calibration through averaging and scaling.
result Improvement in the calibration of phone likelihoods through averaging and scaling of log likelihoods.
Improved Naive Bayes for better phone call behavior classification.
problem Noise in mobile phone data affects phone call behavior classification accuracy.
method Improved naive Bayes classifier with behavioral pattern analysis and dynamic noise threshold.
result Our technique improves classification accuracy by 15%.
Paper uses UKS to improve BLE RSSI for proximity inference in mobile phone apps.
problem Improper BLE RSSI for accurate proximity inference during pandemics.
method Single-dimensional Unscented Kalman Smoother (UKS) with Gaussian process observation transforms.
result UKS outperforms traditional methods in predicting infection risk from BLE RSSI.
Survey of mobility studies using mobile phone data.
problem Understanding human mobility patterns.
method Data Science techniques applied to mobile phone datasets.
result Applications in urban planning, data traffic prediction, etc.
Paper presents a new time-series segmentation technique for mobile phone user behavior.
problem Current segmentation techniques do not accurately capture individual user behavior over time.
method Behavior-Oriented Time Segmentation (BOTS) technique that considers temporal coverage and number of incidences.
result BOTS technique better captures user behavior at various times of day and week.
Mobile phone data predicts users' income levels.
problem Predicting users' income levels from mobile phone data.
method Comparison of feature extraction methods and machine learning techniques.
result Bayesian method based on the communication graph outperforms other methods.
Mobile phone data predicts loan repayment risk.
problem Lack of formal financial histories hinders credit extension.
method Behavioral signatures in mobile phone usage data.
result Mobile phone data outperforms traditional credit scoring methods.
Model predicts cognitive health risks based on smartphone usage patterns.
problem Identifying cognitive health risks through smartphone usage.
method Structured models of smartphone interactions analyzed over 12 weeks.
result AUROC of 0.79 in discriminating between healthy and symptomatic subjects.
Study uses mobile phone data to map Chagas disease risk zones.
problem Identifying geographical spread of Chagas disease.
method Analyzing geolocalized call records and public health information.
result Generated risk maps for public health campaigns.
Improved contact tracing models outperform NIST challenge results.
problem Contact tracing using phone data and machine learning.
method Developed two machine learning models (GBM and MLP) from phone instrumental data features.
result Outperformed the leading NIST challenge result by HKUST.
Develops efficient algorithms for data science, tackling the curse of dimensionality.
problem Tackles the curse of dimensionality in large datasets.
method Focuses on feature extraction techniques and meta-heuristic algorithms, including evolutionary algorithms.
result Evolutionary algorithms are effective in solving optimization problems with a curse of dimensionality.
StrokeSave app diagnoses stroke via mobile sensors and AI.
problem Inadequate stroke diagnosis in areas without quality medical facilities.
method Mobile app with sensors, facial and retinal images, deep learning model.
result Deep learning model achieves 95% accuracy in diagnosing stroke.
Improved phone classification accuracy using graph-based regularization.
problem Phone classification with limited labeled data.
method Graph-based semi-supervised learning with stochastic entropic regularization.
result Significantly improved phone classification accuracy with low labeled data.
The study compares prepaid and postpaid mobile phone users and predicts their subscription type.
problem Predicting mobile phone subscription type based on usage and network connections.
method Graph labelling approach using max-flow min-cut algorithms and indirect inference methods.
result Graph labelling approach achieves 87% classification accuracy, outperforming supervised learning methods.
Improved phone classification using semi-supervised learning with autoencoders.
problem Phone classification accuracy with limited labeled data.
method Semi-supervised learning with sparse autoencoders, using both labeled and unlabelled data.
result The method outperforms standard supervised training and provides competitive error rates.
Predicts customer call intent for auto dealerships using CNN.
problem Understanding customer intent from phone calls for better service.
method Developed a CNN-based supervised learning model for multi-class classification.
result CNN model performs well on customer call intent classification.
Improved topic modeling captures temporal relationships in speech.
problem Lack of temporal information in LDA for speech analysis.
method Temporal Markov chain extension to LDA for acoustic unit discovery.
result Improved phone segmentation results compared to base LDA.
Bayesian approach detects changepoints with cost-sensitive data fidelity.
problem Detecting abrupt shifts in time series data with limited resources.
method Bayesian approach with active, cost-sensitive data fidelity switching.
result Information-based approach reduces total cost while maintaining accuracy.
Direct acoustics-to-word models improve speech recognition without LMs.
problem Improving speech recognition without requiring a Language Model (LM).
method Direct acoustics-to-word CTC models trained on public benchmark tasks.
result CTC word model achieves 13.0%/18.8% word error rate compared to 9.6%/16.0% for phone-based CTC with a 4-gram LM.
Predicts gender and age from mobile phone data for marketing.
problem Enhance marketing offers by predicting customer demographics.
method Machine learning algorithms applied to CDRs, CRM, and billing info.
result 85.6% accuracy in gender prediction, 65.5% in age prediction.
The paper optimizes neural network inference on mobile GPUs.
problem Limited computing power and thermal constraints on mobile CPUs.
method Leverage mobile GPUs for neural network inference.
result Real-time inference of deep neural networks on Android and iOS devices.
This work builds a sensor graph from DC sensors for anomaly detection.
problem Anomaly detection in data centers with complex sensor relationships.
method Data-driven pipeline (ts2graph) to build a sensor graph from sensor time series.
result Graph neural network (GNN) outperforms existing methods by 2-3 times in anomaly detection.
The paper analyzes how speech enhancement and recognition can be improved in noisy environments.
problem Improving speech recognition in multi-talker scenarios with limited resources.
method Developed and trained two LSTM-based models for speech enhancement and phone recognition, then studied their joint optimization.
result Joint optimization of speech enhancement and recognition leads to a significant reduction in Phone Error Rate (PER).
Develops a graph-based convolutional network for multi-view networks to improve poverty research.
problem Binary treatment of social network relations in graph learning models.
method Multi-GCN: Graph Convolutional Networks for Multi-View Networks.
result Multi-GCN outperforms state-of-the-art algorithms on poverty prediction tasks and broader multi-view network tasks.
Proposes E-MIIM for better mobile phone interruptions management.
problem Improper phone interruptions in daily life activities.
method Ensemble learning based on context-aware mobile telephony model.
result E-MIIM outperforms existing MIIM models in prediction accuracy.
Real-time drowsiness detection on mobile phones reduces road trauma.
problem Driver drowsiness increases crash risk and road trauma.
method Depthwise separable 3D convolutions combined with early fusion of spatial and temporal information.
result Real-time drowsiness detection on mobile phones reduces road trauma.
Develops efficient algorithms for spatial field reconstruction and sensor selection in heterogeneous weather sensor networks.
problem Efficient spatial field reconstruction and query-based sensor set selection in heterogeneous sensor networks.
method Spatial Best Linear Unbiased Estimator (S-BLUE) and Cross Entropy method.
result Efficient algorithms with performance guarantees for spatial field reconstruction and sensor selection.
Machine learning models in physical world are vulnerable to subtle adversarial examples.
problem Vulnerability of machine learning models to adversarial examples in physical world.
method Demonstrated vulnerability by feeding adversarial images from cell-phone camera to an ImageNet Inception classifier.
result A large fraction of adversarial examples are classified incorrectly even through a camera.
DAFL learns efficient neural networks without training data.
problem Training data unavailable for deep networks.
method Generative adversarial networks (GANs) to create training samples.
result Achieves high accuracy (92.22%) on CIFAR-10 dataset.
We have recently shown that deep Long Short-Term Memory (LSTM) recurrent neural networks (RNNs) outperform feed forward deep neural networks (DNNs) as acoustic models for speech recognition. More recently, we have shown that the performance of sequence trained context dependent (CD) hidden Markov model (HMM) acoustic m…
Improved robustness in multi-modal sensor fusion with deep learning.
problem Inconsistency in fusion weights leading to poor performance under sensor failures.
method Proposes deep multi-modal sensor fusion architectures with fusion weight regularization and target learning.
result Proposed architectures outperform existing deep learning methods under sensor failures.
Proposes a neural network for handling multi-sensor time series with varying input dimensions.
problem Handling multi-sensor time series with varying input dimensions.
method Graph neural network conditioning vectors for zero-shot transfer learning.
result Better generalization in activity recognition and equipment prognostics datasets.
RESPIRE calibrates low-cost air-quality sensors for CO levels, resistant to outliers.
problem Calibrating LCAQ sensors against regulatory-grade monitors is expensive and time-consuming.
method PROvably outlier-resistant semi-parametric regression technique.
result RESPIRE offers improved prediction in cross-site, cross-season, and cross-sensor settings.