Multiple task transfer learning with small sample sizes

Prognosis, such as predicting mortality, is common in medicine. When confronted with small numbers of samples, as in rare medical conditions, the task is challenging. We propose a framework for classification with data with small numbers of samples. Conceptually, our solution is a hybrid of multi-task and transfer learning, employing data samples from source tasks as in transfer learning, but considering all tasks together as in multi-task learning. Each task is modelled jointly with other related tasks by directly augmenting the data from other tasks. The degree of augmentation depends on the task relatedness and is estimated directly from the data. We apply the model on three diverse real-world data sets (healthcare data, handwritten digit data and face data) and show that our method outperforms several state-of-the-art multi-task learning baselines. We extend the model for online multi-task learning where the model parameters are incrementally updated given new data or new tasks. The novelty of our method lies in offering a hybrid multi-task/transfer learning model to exploit sharing across tasks at the data-level and joint parameter learning.

Prosthetic motor imaginary task classification using single channel of electroencephalography

Brain Computer Interface (BCI) is playing a very important role in human machine communications. Recent communication systems depend on the brain signals for communication. In these systems, users clearly manipulate their brain activity rather than using motor movements in order to generate signals that could be used to give commands and control any communication devices, robots or computers. In this paper, the aim was to estimate the performance of a brain computer interface (BCI) system by detecting the prosthetic motor imaginary tasks by using only a single channel of electroencephalography (EEG). The participant is asked to imagine moving his arm up or down and our system detects the movement based on the participant brain signal. Some features are extracted from the brain signal using Mel-Frequency Cepstrum Coefficient and based on these feature a Hidden Markov model is used to help in knowing if the participant imagined moving up or down. The major advantage in our method is that only one channel is needed to take the decision. Moreover, the method is online which means that it can give the decision as soon as the signal is given to the system. Hundred signals were used for testing, on average 89 % of the up down prosthetic motor imaginary tasks were detected correctly. This method can be used in many different applications such as: moving artificial prosthetic limbs and wheelchairs due to it's high speed and accuracy.

Prosthetic motor imaginary task classification based on EEG quality assessment features

Brain Computer Interface (BCI) plays an important role in the communication between human and machines. This communication is based on the human brain signals. In these systems, users use their brain instead of the limbs or body movements to do tasks. The brain signals are analyzed and translated into commands to control any communication devices, robots or computers. In this paper, the aim was to enhance the performance of a brain computer interface (BCI) systems through better prosthetic motor imaginary tasks classification. The challenging part is to use only a single channel of electroencephalography (EEG). Arm movement imagination is the task of the user, where (s)he was asked to imagine moving his arm up or down. Our system detected the imagination based on the input brain signal. Some EEG quality features were extracted from the brain signal, and the Decision Tree was used to classify the participant's imagination based on the extracted features. Our system is online which means that it can give the decision as soon as the signal is given to the system (takes only 20 ms). Also, only one EEG channel is used for classification which reduces the complexity of the system which leads to fast performance. Hundred signals were used for testing, on average 97.4% of the up-down prosthetic motor imaginary tasks were detected correctly. This method can be used in many different applications such as: moving artificial limbs and wheelchairs due to it's high speed and accuracy.

Collaborating differently on different topics: a multi-relational approach to multi-task learning

Multi-task learning offers a way to benefit from synergy of multiple related prediction tasks via their joint modeling. Current multi-task techniques model related tasks jointly, assuming that the tasks share the same relationship across features uniformly. This assumption is seldom true as tasks may be related across some features but not others. Addressing this problem, we propose a new multi-task learning model that learns separate task relationships along different features. This added flexibility allows our model to have a finer and differential level of control in joint modeling of tasks along different features. We formulate the model as an optimization problem and provide an efficient, iterative solution. We illustrate the behavior of the proposed model using a synthetic dataset where we induce varied feature-dependent task relationships: positive relationship, negative relationship, no relationship. Using four real datasets, we evaluate the effectiveness of the proposed model for many multi-task regression and classification problems, and demonstrate its superiority over other state-of-the-art multi-task learning models

Prediciton of emergency events: a multi-task multi-label learning approach

Prediction of patient outcomes is critical to plan resources in an hospital emergency department. We present a method to exploit longitudinal data from Electronic Medical Records (EMR), whilst exploiting multiple patient outcomes. We divide the EMR data into segments where each segment is a task, and all tasks are associated with multiple patient outcomes over a 3, 6 and 12 month period. We propose a model that learns a prediction function for each task-label pair, interacting through two subspaces: the first subspace is used to impose sharing across all tasks for a given label. The second subspace captures the task-specific variations and is shared across all the labels for a given task. The proposed model is formulated as an iterative optimization problems and solved using a scalable and efficient Block co-ordinate descent (BCD) method. We apply the proposed model on two hospital cohorts - Cancer and Acute Myocardial Infarction (AMI) patients collected over a two year period from a large hospital emergency department. We show that the predictive performance of our proposed models is significantly better than those of several state-of-the-art multi-task and multi-label learning methods.

A new transfer learning framework with application to model-agnostic multi-task learning

Learning from small number of examples is a challenging problem in machine learning. An effective way to improve the performance is through exploiting knowledge from other related tasks. Multi-task learning (MTL) is one such useful paradigm that aims to improve the performance through jointly modeling multiple related tasks. Although there exist numerous classification or regression models in machine learning literature, most of the MTL models are built around ridge or logistic regression. There exist some limited works, which propose multi-task extension of techniques such as support vector machine, Gaussian processes. However, all these MTL models are tied to specific classification or regression algorithms and there is no single MTL algorithm that can be used at a meta level for any given learning algorithm. Addressing this problem, we propose a generic, model-agnostic joint modeling framework that can take any classification or regression algorithm of a practitioner’s choice (standard or custom-built) and build its MTL variant. The key observation that drives our framework is that due to small number of examples, the estimates of task parameters are usually poor, and we show that this leads to an under-estimation of task relatedness between any two tasks with high probability. We derive an algorithm that brings the tasks closer to their true relatedness by improving the estimates of task parameters. This is achieved by appropriate sharing of data across tasks. We provide the detail theoretical underpinning of the algorithm. Through our experiments with both synthetic and real datasets, we demonstrate that the multi-task variants of several classifiers/regressors (logistic regression, support vector machine, K-nearest neighbor, Random Forest, ridge regression, support vector regression) convincingly outperform their single-task counterparts. We also show that the proposed model performs comparable or better than many state-of-the-art MTL and transfer learning baselines.

Computational intelligence-based traffic signal timing optimization

 Traffic congestion has explicit effects on productivity and efficiency, as well as side effects on environmental sustainability and health. Controlling traffic flows at intersections is recognized as a beneficial technique, to decrease daily travel times. This thesis applies computational intelligence to optimize traffic signals' timing and reduce urban traffic.

What shall i share and with whom? A multi-task learning formulation using multi-faceted task relationships

Multi-task learning is a learning paradigm that improves the performance of "related" tasks through their joint learning. To do this each task answers the question "Which other task should I share with"? This task relatedness can be complex - a task may be related to one set of tasks based on one subset of features and to other tasks based on other subsets. Existing multi-task learning methods do not explicitly model this reality, learning a single-faceted task relationship over all the features. This degrades performance by forcing a task to become similar to other tasks even on their unrelated features. Addressing this gap, we propose a novel multi-task learning model that leams multi-faceted task relationship, allowing tasks to collaborate differentially on different feature subsets. This is achieved by simultaneously learning a low dimensional sub-space for task parameters and inducing task groups over each latent subspace basis using a novel combination of L1 and pairwise L∞ norms. Further, our model can induce grouping across both positively and negatively related tasks, which helps towards exploiting knowledge from all types of related tasks. We validate our model on two synthetic and five real datasets, and show significant performance improvements over several state-of-the-art multi-task learning techniques. Thus our model effectively answers for each task: What shall I share and with whom?

Design of distributed adaptive neural traffic signal timing controller by cuckoo search optimization

This paper focuses on designing an adaptive controller for controlling traffic signal timing. Urban traffic is an inevitable part in modern cities and traffic signal controllers are effective tools to control it. In this regard, this paper proposes a distributed neural network (NN) controller for traffic signal timing. This controller applies cuckoo search (CS) optimization methods to find the optimal parameters in design of an adaptive traffic signal timing control system. The evaluation of the performance of the designed controller is done in a multi-intersection traffic network. The developed controller shows a promising improvement in reducing travel delay time compared to traditional fixed-time control systems.

Simulated firefighting task performance and physiology under very hot conditions

PURPOSE: To assess the impact of very hot (45°C) conditions on the performance of, and physiological responses to, a simulated firefighting manual-handling task compared to the same work in a temperate environment (18°C). METHODS: Ten male volunteer firefighters performed a 3-h protocol in both 18°C (CON) and 45°C (VH). Participants intermittently performed 12 × 1-min bouts of raking, 6 × 8-min bouts of low-intensity stepping, and 6 × 20-min rest periods. The area cleared during the raking task determined work performance. Core temperature, skin temperature, and heart rate were measured continuously. Participants also periodically rated their perceived exertion (RPE) and thermal sensation. Firefighters consumed water ad libitum. Urine specific gravity (USG) and changes in body mass determined hydration status. RESULTS: Firefighters raked 19% less debris during the VH condition. Core and skin temperature were 0.99 ± 0.20 and 5.45 ± 0.53°C higher, respectively, during the VH trial, and heart rate was 14-36 beats.min(-1) higher in the VH trial. Firefighters consumed 2950 ± 1034 mL of water in the VH condition, compared to 1290 ± 525 in the CON trial. Sweat losses were higher in the VH (1886 ± 474 mL) compared to the CON trial (462 ± 392 mL), though both groups were hydrated upon protocol completion (USG < 1.020). Participants' average RPE was higher in the VH (15.6 ± 0.9) compared to the CON trial (12.6 ± 0.9). Similarly, the firefighers' thermal sensation scores were significantly higher in the VH (6.4 ± 0.5) compared to the CON trial (4.4 ± 0.4). CONCLUSIONS: Despite the decreased work output and aggressive fluid replacement observed in the VH trial, firefighters' experienced increases in thermal stress, and exertion. Fire agencies should prioritize the health and safety of fire personnel in very hot temperatures, and consider the impact of reduced productivity on fire suppression efforts.

Toxicity prediction in cancer using multiple instance learning in a multi-task framework

Treatments of cancer cause severe side effects called toxicities. Reduction of such effects is crucial in cancer care. To impact care, we need to predict toxicities at fortnightly intervals. This toxicity data differs from traditional time series data as toxicities can be caused by one treatment on a given day alone, and thus it is necessary to consider the effect of the singular data vector causing toxicity. We model the data before prediction points using the multiple instance learning, where each bag is composed of multiple instances associated with daily treatments and patient-specific attributes, such as chemotherapy, radiotherapy, age and cancer types. We then formulate a Bayesian multi-task framework to enhance toxicity prediction at each prediction point. The use of the prior allows factors to be shared across task predictors. Our proposed method simultaneously captures the heterogeneity of daily treatments and performs toxicity prediction at different prediction points. Our method was evaluated on a real-word dataset of more than 2000 cancer patients and had achieved a better prediction accuracy in terms of AUC than the state-of-art baselines.

Differentially private multi-task learning

Privacy restrictions of sensitive data repositories imply that the data analysis is performed in isolation at each data source. A prime example is the isolated nature of building prognosis models from hospital data and the associated challenge of dealing with small number of samples in risk classes (e.g. suicide) while doing so. Pooling knowledge from other hospitals, through multi-task learning, can alleviate this problem. However, if knowledge is to be shared unrestricted, privacy is breached. Addressing this, we propose a novel multi-task learning method that preserves privacy of data under the strong guarantees of differential privacy. Further, we develop a novel attribute-wise noise addition scheme that significantly lifts the utility of the proposed method. We demonstrate the effectiveness of our method with a synthetic and two real datasets.

Functional timing analysis of VLSI circuits containing complex gates

The recent advances in CMOS technology have allowed for the fabrication of transistors with submicronic dimensions, making possible the integration of tens of millions devices in a single chip that can be used to build very complex electronic systems. Such increase in complexity of designs has originated a need for more efficient verification tools that could incorporate more appropriate physical and computational models. Timing verification targets at determining whether the timing constraints imposed to the design may be satisfied or not. It can be performed by using circuit simulation or by timing analysis. Although simulation tends to furnish the most accurate estimates, it presents the drawback of being stimuli dependent. Hence, in order to ensure that the critical situation is taken into account, one must exercise all possible input patterns. Obviously, this is not possible to accomplish due to the high complexity of current designs. To circumvent this problem, designers must rely on timing analysis. Timing analysis is an input-independent verification approach that models each combinational block of a circuit as a direct acyclic graph, which is used to estimate the critical delay. First timing analysis tools used only the circuit topology information to estimate circuit delay, thus being referred to as topological timing analyzers. However, such method may result in too pessimistic delay estimates, since the longest paths in the graph may not be able to propagate a transition, that is, may be false. Functional timing analysis, in turn, considers not only circuit topology, but also the temporal and functional relations between circuit elements. Functional timing analysis tools may differ by three aspects: the set of sensitization conditions necessary to declare a path as sensitizable (i.e., the so-called path sensitization criterion), the number of paths simultaneously handled and the method used to determine whether sensitization conditions are satisfiable or not. Currently, the two most efficient approaches test the sensitizability of entire sets of paths at a time: one is based on automatic test pattern generation (ATPG) techniques and the other translates the timing analysis problem into a satisfiability (SAT) problem. Although timing analysis has been exhaustively studied in the last fifteen years, some specific topics have not received the required attention yet. One such topic is the applicability of functional timing analysis to circuits containing complex gates. This is the basic concern of this thesis. In addition, and as a necessary step to settle the scenario, a detailed and systematic study on functional timing analysis is also presented.

Improving mutual fund market timing measures: a markov switching approach

Market timing performance of mutual funds is usually evaluated with linear models with dummy variables which allow for the beta coefficient of CAPM to vary across two regimes: bullish and bearish market excess returns. Managers, however, use their predictions of the state of nature to deÞne whether to carry low or high beta portfolios instead of the observed ones. Our approach here is to take this into account and model market timing as a switching regime in a way similar to Hamilton s Markov-switching GNP model. We then build a measure of market timing success and apply it to simulated and real world data.