Machine Learning for Healthcare Applications. Группа авторов
Читать онлайн книгу.
is basically a reading of brain’s electric voltage fluctuations as read on scalp electrodes. It is the approximate cumulative electrical activity of neurons. This process is one of the best non-penetrative interfaces because of its temporal resolution. But it has hindrances like susceptibility to noise which is a very prominent barrier to implementing EEG devices as BCI solutions. It requires extensive training for users and models to provide substantial results in a consistent manner.
As for example, Niels Birbaumer from University of Tubingen had brought paralyzed patients in mid-1990s for training them to control the slow cortical potentials to be utilized in order for them to control a computer’s cursor by binary signaling. They were slow, as in they required 1 h to write 100 characters and training them took several months but it appeared as a breakthrough possibility.
3.1.4 History of EEG
Hans Berger was the man who discovered that there is significant electrical activity in the brain and developed the initial process known as Electroencephalography today. In the year 1924 he captured the first brain signal and by analyzing them he found oscillatory activity in the form of alpha wave (8–12 Hz) called Berger’s wave. During early days he inserted silver wires in the scalp of patients then after graduating to using silver foils which in turn were connected to a “Lippmann Capillary Electrometer”. Later on he experimented with galvanometers and after significant analysis he started creating brain maps of electrical pulses for specific brain diseases. This all led to the discovery of EEG and created new possibilities in human advancement.
3.1.5 About Neuromarketing
Generally marketing procedures or tools include surveying, interviews, target groups, etc. where participants willingly give feedback for their thoughts and opinions on a product. These procedures have a drawback of not considering the unconscious decision-making characteristics. EEG has the potential to identify these emotions and influence the decision-making process. It is found that 90% of the decision making consists of factors from the subconscious mind. In this domain, the analysis of the subconscious mind presents the true choices of users more accurately than other methods. It has the potential to factor in characteristics about decision making in users which cannot be accurately pin-pointed in other methods. Neuromarketing fills the gap between the results of traditional marketing methods and real decisions of consumers.
In order to analyze the consumer’s behavior we factor in his sensorimotor and mental feedback with the combination of eye-tracking, skin conductance, galvanic skin response and facial electromyography which all in turn contribute to create a sequential flow chart of a consumer’s response and various stimuli which resulted in the failure or successful of purchase of the product.
This technology came around prominence in 2002 with the initiatives of Brighthouse and SalesBrain who developed marketing solutions based on Neuromarketing and also offered studies on the client’s product base from a consumer’s EEG data. Thus, major powerhouses in FMCG markets have started to adopt Neuromarketing techniques to exploit the consumer’s behaviour from a marketing and psychological angle.
3.1.6 About Machine Learning
“Machine Learning” is the process of enabling computing devices to try, learn, do and verify assigned tasks to be performed on their own without being hard-coded to do so. The learning process in the code needs to evolve on its own with the changing parameters and perform accordingly. Earlier on human used to analyze all the scenarios in a task and we would dictate the steps required to the computer but during wider and more complex situations we realized it’s better for the computers to develop its own algorithm.
This domain of Machine Learning implements various tracks to help computers automatedly accomplish tasks where no correct sequence of steps is known. It involves training the computer through a vast number of situations and label them as successful or not and accordingly perform the correct sequence of action for the respective situation. This is called training data which is used to improve the effectiveness and efficiency of an algorithm. We have extensively used this procedure to achieve significant results in regard to our efficiency in predicting the choice of a consumer and also establishing a brain map for like/dislike.
3.2 Literature Survey
In this section, we shall briefly mention most of the studies and research done currently in the field of EEG and Neuromarketing relevant to our study.
Primarily we got a solid foundation [1] to work on from Dr. Partha Pratim Roy’s and his associates’ paper on Analysis of EEG signals and application of Neuromarketing, in his paper he has used the deep learning method of Hidden Markov Model and recorded the dataset using user-independent test approach. He also proposed a predictive modelling framework to acquire the consumer’s knowledge about what all he like/dislikes amongst the sample products using an Emotiv EPOC+ sensor. We have borrowed his dataset for initial study as an ice-breaker and it has helped us in leaps.
After reading his paper, we inherently searched for the spectrum of mind which consciously makes the decision of a person liking/disliking a product in a natural environment. We had encountered lot of reasons such as presentation, composition of materials, past experiences, cost and brand value which a person uses to determine its likeability. But perhaps this wasn’t enough. So, we decided to delve a little into emotion recognition for identifying which all areas in brain elicit an emotion. Following will be our concise notes on emotion recognition and after which we shall provide the methodological research of models.
This paper [2] is about automatic emotional classification by EEG data using DEAP dataset led by Samarth Tripathi and his associates, applying Convolutional and Deep Neural Networks on DEAP datasets. Earlier emotion recognition involved text, speech, facial, etc. as analyzing parameter s. An emotion is a psychophysiological operation started by a voluntary or involuntary reception of a situation.
In this paper, peripheral physiological signals of 32 subjects were recorded while they watched videos and were evaluated on levels of arousals & valence. They used a 32 EEG-channel 512 Hz Biosemi Active2 device that utilizes active AgCl electrodes to compile the data.
Neural networks implement functions based on large datasets of unknown inputs by training & statistic models. Here, 2 neural models are used 1. Deep Neural Network (DNN) and 2. Convolutional Neural Network (CNN). The dataset is of 8064 signal data from 40 channels for each subject. A total of 322,560 readings were recorded for the models to process. The first model, i.e., DNN used 4 neural levels whose output became input for the subsequent levels. As the dataset was limited, they implemented dropout technique with superior Epoch which could keep the count of all training vectors for updating weights. The datasets were divided into groups for easier use and they all go through learning algorithms before Epoch update occurs. The data was trained in 310 groups with Epoch of 250. For the second model, i.e., CNN, the DEAP data is converted to 2D images for the 101 readings each totaling to a size of 4,040 units. CNN’s first layer used ‘Tan Hyperbolic’ as activation function in valence classification model & ‘Relu’ as activation for arousal model. The subsequent levels used 100 filters and 3 ∗ 3 sized kernel with the very same ‘Tan Hyperbolic’ function as activation function for both classifier models. The last dense layer used ‘Softplus’ as its activation function using CCE as loss function and SGD as an optimizer.
The learning rates were found to be 0.00001 for valence, 0.001 for arousal & a gradient momentum of 0.9. These models resulted in 4.51 & 4.96% improvement in classifying valence and arousal respectively among 2 classes (High/Low) in valence & 3 classes (High/Normal/Low) in arousal. The learning rate is marginally more useful, but dropout probability secures the best classification across levels. They also noted that wrong choice of activation functions especially 1st CNN layer will cause severe defects to models. The models were highly accurate with respect to previous researchers and prove the fact that neural networks are the key for EEG classification of emotions in a step to unlocking the brain.
Hence Deep Neural Networks are used to analyze human emotions and classify them by PSD and