Emotion_Recognition_using_Brainwave_Datasets_Bethlehem_Seid (1).pptx
- 1. Emotion Recognition using Brainwave Datasets Heba Elgazzar and Bethlehem Seid School of Engineering and Computer Science Morehead State University Morehead, KY, USA
- 2. Outline 1. Introduction and Related Work 2. Proposed Method 3. Machine Learning Algorithms 4. Results 5. Conclusion
- 3. Introduction ā¢ Traditionally, we use speech as a form of communication when we are feeling depressed or anxious. ā¢ Expressions are vague and donāt describe our emotions very well. ā¢ Therefore, it is essential to find other methods to understand the mental state of an individual. ā¢ Our bodies uses electric signals to send and receive messages so measuring the electrical signals to see the mental state of individuals
- 4. Introduction ā¢ The dataset used for this research limits us to 3 emotional state which is positive, negative, and neutral. ā¢ But this can be easily expanded to other emotions. ā¢ This EEG test was also expensive and required the assistance of medical professionals to interpret the meaning of the signals. ā¢ However, with the improvements in technology low-cost commercial devices such as the Emotiv, NeuroSky, and MUSE headband the process of collecting brainwave dataset and interpreting the results has been simplified.
- 5. Related Work ā¢ Other researchers have worked on similar topics using EEG headbands ā¢ Detecting mental state relaxed, concentrated, and neutral ā¢ Detecting different positive emotions ā¢ Detecting emotional state using EEG headband
- 6. Proposed Method ā¢ MUSE EEG headband was used ā¢ The headband had four electrodes placed on the TP9, AF7, AF8 and TP10 ā¢ Used to capture the readings within a time window.
- 7. Dataset ā¢ We used an existing public EEG Brainwave dataset this was described and released in for classifying emotional state. ā¢ In this dataset 2548 different attributes and 2133 instances ā¢ Two individuals one male and one female were shown 6 different scenes to invoke negative and positive emotions. Three minutes of each state was recorded. ā¢ In addition, 6 minutes of resting neutral state was recorded.
- 8. Dataset To invoke negative emotions ā¢ Death scene from Marley and Me produced by Twentieth Century studios ā¢ Opening death scene from Up produced by Walt Disney Pictures ā¢ Funeral scene from My Girl produced by Imagine Entertainment. To invoke positive emotions ā¢ Opening musical number from La La Land produced by Summit Entertainment ā¢ Nature time-lapse from slow life by BioQuest Studios, ā¢ Funny dog clips by MashupZone
- 9. ā¢ Anaconda which is a free and open-source platform used for scientific computation and package management. ā¢ Pandas which is a software library for python programming and helps with the data manipulation and analysis. ā¢ Matplotlib which is a library for plotting
- 10. ā¢ Supervised machine learning techniques were chosen for this research ā¢ Supervised learning is where a dataset and its labels is provided. ā¢ The data is then split into training and testing sets ā¢ Training set is used to develop algorithm can produce a way to predict the label of an unknown item by using the information previously provided. ā¢ Testing set is used to measure how well the algorithm is performing by comparing the predicted label to the actual label given in the dataset. Dataset
- 11. The classification algorithm used in this research are: ā¢ K-Nearest Neighbors(KNN) ā¢ Decision Trees ā¢ Random Forest ā¢ Artificial Neural Network(ANN). Dataset
- 12. K-nearest neighbors KNN ā¢ KNN is a simple machine learning algorithm ā¢ Based on data elements that are classified in a group will have similar attributes. ā¢ So, a data will be classified into a group by first finding the closest points then classify the point based on majority vote of the K neighbors. ā¢ The closeness will be calculated using the Euclidean distance ššš š” š„, š¦ = š=1 š (š„1 ā š¦1)2 0 2 4 6 8 10 12 0 2 4 6 8 10 12
- 13. Pros ā¢ Training is much faster compared to other classification algorithms. ā¢ Useful in case of nonlinear data. ā¢ Used with the regression problem. Cons ā¢ Slow and costly ā¢ Requires large memory for storing the entire training dataset for prediction. ā¢ Not suitable for large dimensional data. K-nearest neighbors KNN
- 14. Decision Trees ā¢ Flowchart-like upside-down tree structure. ā¢ Root represents the entire dataset ā¢ Nodes represent a single feature ā¢ Branch represents the decision rule for partitioning ā¢ Leaves represent the outcome/classes. ā¢ Partitioning is done by first calculating the entropy which is the measure of impurity
- 15. Decision Trees Pros ā¢ Decision trees are easy to interpret and visualize. ā¢ It can easily capture Non-linear patterns. ā¢ The decision tree has no assumptions about distribution Cons ā¢ Sensitive to noisy data. ā¢ The small variation in data can result in the different decision tree ā¢ Decision trees are biased
- 16. Random Forest ā¢ Similar to decision tree ā¢ Overcomes the drawbacks of the decision tree ā¢ Creates several decision trees ā¢ It classifies the data based on majority vote from the several decision trees
- 17. Artificial Neural Networks ā¢ Modeled after the human brain ā¢ Processes information in a very complex and nonlinear fashion by using a web of neurons that communicate through electric signals. ā¢ If the sum reaches a certain threshold, then it transmits message to the other neurons. ā¢ Here, weight represents the strength of the connection of between the two neurons. ā¢ Weight and the bias are the parts that are adjusted as the neural network is learning.
- 18. Feature Selection Principal Component Analysis (PCA) ā¢ Linear dimensionality reduction ā¢ Preserves the essential parts that have more variation of the data ā¢ Cluster the similar data points based on the feature correlation ā¢ Features selected based on the variance or based on the number of components 0 2 4 6 8 10 12 0 2 4 6 8 10 12 Chart Title -8 -6 -4 -2 0 2 4 6 8 0 2 4 6 8 10 12 14 Chart Title
- 19. Feature Selection SelectKBest ā¢ The correlation of the different independent features and the outcome ā¢ Univariant feature selection ā¢ Pick a feature do a chi-square test with the target variable find out the statistical significance of the test of independence and keep feature based on the test of the significance then select the top K features to run the data on ReliefF ā¢ Filter-method approach to feature selection ā¢ Scoring is based on the identifying the featureās value difference compared to its nearest neighbor ā¢ The closest same-class instance is called 'near-hit', and the closest different-class instance is called 'near-miss'
- 20. Results Using PCA 89.1% 92.51% 95.55% 95.66% 84 86 88 90 92 94 96 98 KNN Decision Tree Random Forest ANN Accuracy Machine Learning Classifiers Accuracy of diffrent machine learning algorithms
- 21. Results Using SeleckKBest 75.78% 85.9% 89% 81.25% 65 70 75 80 85 90 KNN Decision Tree Random Forest ANN Accuracy Machine Learning Classifiers Accuracy of diffrent machine learning algorithms
- 22. Conclusion ā¢ In this research, we proposed a novel method for classifying the emotional state of individuals using few features. ā¢ ANN shows promising results when using principal component analysis ā¢ Random forest performed better when using SelectKBest ā¢ KNN showed the least accurate results
- 23. Thank You
Editor's Notes
- #24: import matplotlib.pyplot as plt import numpy as np n_groups = 4 means_frank = (89.1, 92.51, 95.55, 95.66) fig, ax = plt.subplots() index = np.arange(n_groups) bar_width = 0.4 opacity = 0.8 rects1 = plt.bar(index, means_frank, bar_width,alpha=opacity,color='red') plt.xlabel('') plt.ylabel('Accuracy') plt.title('Accuracy of diffrent classifier') plt.xticks(index, ('KNN', 'Decision Tree', 'Random Forest', 'ANN')) plt.legend() plt.tight_layout() plt.show() import matplotlib.pyplot as plt import numpy as np n_groups = 4 means_frank = (75.78, 85.9, 89, 81.25) fig, ax = plt.subplots() index = np.arange(n_groups) bar_width = 0.4 opacity = 0.8 rects1 = plt.bar(index, means_frank, bar_width,alpha=opacity,color='red') plt.xlabel('') plt.ylabel('Accuracy') plt.title('Accuracy of diffrent classifier') plt.xticks(index, ('KNN', 'Decision Tree', 'Random Forest', 'ANN')) plt.legend() plt.tight_layout() plt.show() import matplotlib.pyplot as plt import numpy as np n_groups = 4 means_frank = (75.78, 85.9, 89, 81.25) fig, ax = plt.subplots() index = np.arange(n_groups) bar_width = 0.4 opacity = 0.8 rects1 = plt.bar(index, means_frank, bar_width,alpha=opacity,color='red') plt.xlabel('') plt.ylabel('Accuracy') plt.title('Accuracy of diffrent classifier') plt.xticks(index, ('KNN', 'Decision Tree', 'Random Forest', 'ANN')) plt.legend() plt.tight_layout() plt.show()