Project

Question 1

USP of our paper

Answer 1

o Identification of vivax parasite: little research but most diversified geographically & no vaccine exists yet on market <-> falciparum
o Zhao et al. (2020) tested falciparum trained model to identified vivax -> overfitting
o Binary classification task with this dataset (others only multi) -> because info about malaria existence most important with high accuracy (stage can be identified next) -> results show that CNN1 model same or outperformed multilabel classificaiton with the same BBBC dataset 98,3% (Li et al. 98,3% & Meng et al 94,17%)

Question 2

Importance

Answer 2

rapid diagnosis & treatment best method to prevent severe outcomes & deaths
o No vaccine
o PCR costly & lack of equipment & health personnel
o Lack of health experts in low resource countries for manual microscopy identification
-> ML = cost-efficient, less expert knowledge necessary (as machine detects) & less equipment (no PCR, but just microscope & ml tool)

Question 3

Business Case

Answer 3

o Our business case focuses on leveraging machine learning to improve malaria detection, specifically targeting the P. vivax strain. With over 250 million cases and 600,000 deaths reported annually, malaria remains a significant global health challenge affecting 84 countries. By addressing the neglected P. vivax strain and utilizing advanced machine learning techniques, we aim to provide accurate and cost-effective malaria detection solutions
o Target businesses: clinics (in low resource countries), pharmaceutical companies, public healthcare agencies or NGOs, researchers
o Advantages: cost- & expert-efficient, enhanced diagnosis accuracy & research advancement
o Market: 84 countries effected by malaria (developing countries)

Question 4

Why SVM?

Answer 4

o used in literature similar tasks, compare to other results as well
o machine learning comparison to deep learning
o Ability to handle high-dimensional data using kernel trick
(= random forest, <-> KNN, Naïve Bayes, Linear Models (logisitc & linear regression))
o Robust against overfitting (by tuning C -> regularization parameter deciding size of hyperplane margin, if high -> low training error but lower generalization)
o Effective for binary classification (due to separation by hyperplane)
o Memory efficient (image with many pixel): Only have to save support vectors for decision boundaries (<-> KNN)

Question 5

Why CNN? Why did use CNN and not fully connected network?

Answer 5

o Need to flatten image to 1D vector. Any spatial information is lost! (Position and layout of elements in two-dimensional space), want to identify infected cell part where it doesn’t matter at what location this is exactly
o Solution: convolutional layers -> only partially connect previous layer -> reduced number of connections needed
o Ability learn & identify intricate patterns & structures
o Learn hierarchical representations
o Widely used for image classification

Question 6

Why different CNNs?

Answer 6

o One build from scratch
o One adapted from similar classification task
o Comparing if similar task CNN will also be good for other purpose (ours, vivax parasite, this dataset) -> need for other, specifically trained models?

Question 7

Why scaling between 0 & 1 of the image pixels (normalizing)?

Answer 7

o Equal importance of features: prevents that certain features with large number dominate the learning process (more important for learning) -> biased towards features with larger magnitude
o Enhancing convergence speed (convergence = reaching a stable & optimal state)

Question 8

How did you handle unbalanced dataset?

Answer 8

o Undersampling: randomly choosing 5.000 uninfected images
o Oversampling: ADASYNE

Question 9

What is Data Augmentation?

Answer 9

o Increasing number of training data by creating new samples by applying transformations of existing data
o We used: geometric transformations: rotation & flipping (horizontal & vertical)
o Not: scaling or cropping: losing important details of cell images -> boundary boxes close to each other
o Used: ImageDataGenerator from TensorFlow
o Only for CNN (SVM serves as baseline: beneficial to keep the model as simple as possible to have a clear foundation for comparison with other more advanced approaches) -> could have tried with data augmentation as well

Question 10

Why Data Augmentation?

Answer 10

o more variations and diversity into the training data -> model generalizes better
o mittigating overfitting: larger & more diverse dataset -> learns the characteristics insteady of memorizing training sample -> better generalization

Question 11

Why not use Data Augmentation for SVM?

Answer 11

o Image Generator in Tensorflow not possible to implement in SVM flow
o Baseline model -> should have “true” images

Question 12

What optimizer?

Answer 12

Adam: Adaptive Moment estimation
- Extension of gradient descent
- Combines two algorithms: computes first-oder momentum (moving average of gradients) and second-oder momentum (moving average of squared gradients) of loss function
-> adapts learning rate for each parameter based on its historical gradients & momentum
- GD with momentum = taking into consideration the ‘exponentially weighted average’ of the gradients -> average = converge faster
- RMSP algorithm = taking into consideration the “exponential moving average” (average change in data over time)

Question 13

ADAM - elements

Answer 13

• Adaptive Learning Rate (dynamically adapts it for each parameter in network -> smooth & fast convergence)
• Momentum (helps accelerate convergence by adding a fraction of the previous gradient’s direction to the current gradient update -> overcome local minima)
• Bias correction (to address

Question 14

Why used Adam?

Answer 14

 Used in similar study
 Pro: efficiency & robustness, converge faster, works well on noisy & sparse data
o Difference to normal stochastic GD
 Learning grate adaptive (GD: maintains single learning rate during training)

Question 15

• What is a momentum?

Answer 15

o technique where term is a added to the parameter updates that accounts for previous direction of movement
o e.g momentum value of 0.9: 90% of the previous direction is retained, and only 10% is influenced by the current gradient

Question 16

What are different ways to choose the learning rate in gradient decent?

Answer 16

o Fixed learning rate: constant learning rate is set before training
o Learning rate schedules: adjust learning rate over time
 Step decay: Learning rate is reduced by a certain factor after a fixed number of epochs or iterations
 Exponential decay: The learning rate decreases exponentially over time.
 Performance-Based Schedules: The learning rate is adjusted based on the performance on a validation set
o Adaptive learning rates: dynamically adjust the learning rate based on the behavior of the optimization process.
 !!Adam combines adaptive learning rates with momentum. It adapts the learning rate for each parameter based on both the first-order (gradient) and second-order (gradient squared) moments of the gradients.
o Learning Rate Search: e.g. grid search

Question 17

What does a learning rate of 0.001 mean?

Answer 17

o represents the step size at which a machine learning algorithm updates the model parameters during training = magnitude of adjusments made to model based on calculated gradients
o hyperparameter
o 0.0001 = low learning rate
o Small step sizes, doesn’t overshoot minimum but takes long time to converge

Question 18

What does binary cross entropy loss mean?

Answer 18

o Cost function that measure the difference between the predicted probability distribution and the true probability distribution for a binary classification problem
o calculates the average of the logarithmic loss for each instance, where a higher loss is assigned to incorrect predictions and a lower loss to correct predictions
- loss: H. (q) =-1/N * sum(yi * log(p(yi)) + (1-yi) * log(1-p(yi))
- yi = 1 or 0 true outcome
- p(yi) probability that this outcome
- log: If p = 1 0 -log(p) outcome, if p smaller larger outcome

Question 19

What is overfitting?

Answer 19

performance high on training data but low on validation set (low bias, but high variance)
o observing the learning curves reveals a pattern of overfitting when the training loss consistently decreases, but the validation loss begins to rise or remains stagnant.
o Goal: small gap between training & validation curve -> generalization good

Question 20

What done to prevent overfitting?

Answer 20

• CNN:  Drop out layer, early stopping
• SVM: PCA: removes noise in the data and keeps only the most important features. Less dimensions reduces the risk of overfitting.
   Grid search & cross-validation
   Look at each result of each fold of cross-validation -> if approxi. Same no overfitting

Question 21

how do you choose number of folds (cross-validation)?

Answer 21

o higher number of folds -> more accurate estimate of performance but require more computational resources and time
o Smaller datasets may benefit from a higher number of folds, while larger datasets may be adequately assessed with fewer folds
o We used 3 folds (cv=3) to reduce the CPU time needed (we could have used 5)

Question 22

What is GPU?

Answer 22

o GPU stands for Graphics Processing Unit
- is a specialized electronic circuit designed to quickly process and render graphics
- widely used for accelerating computations in areas such as ml

Question 23

How did you tune hp in CNNs? And which ones?

Answer 23

o With babysitting
o number of hidden layers, units per layer, learning rate, epochs, batch size, early stop patience

Question 24

What are epochs?

Answer 24

o Number of times entire training dataset is passed through NN during training process
o Each epoch consists of: 1 forwards, 1 backward pass (update of weights) for all training examples
o Result: NN can learn & refine parameters gradually

Question 25

What is the activation function ReLu and why does it work well for images?

Answer 25

o Replaces all negative values (pixels) with 0, keep all positive values the same
o ReLu are easy to compute, computationally efficient, gradients are cleanly defined and are constant (except for a piecewise non-linearity)
o introduce non-linearity into the network, allowing it to learn complex patterns and gradients

Question 26

Why PCA only for SVM

Answer 26

 CNN detected weights and features as deep learning
 SVM only weights as machine learning

Question 27

What are kernal functions? SVM

Answer 27

 To classify non linearly sepearble data
 transform the images into a high-dimensional space and consequently find the optimal decision boundaries in this new high-dimensional space
 Functions include: linear, polynomial, radial basis function

Question 28

Which kernal function did we use & why?

Answer 28

 Radial basis function: combines multiple polynomial kernels multiple times of different degrees to project the non-linearly
 Detect via grid search

Question 29

Why grid search to select hyperparameters

Answer 29

• Random Search & Try-Error: needs more time, random and thus expected to provide worse hp, not structured search
• PCA -> less dimensions -> grid search computationally ok

Question 30

What are relevant hyperparameters of SVM?

Answer 30

 C = controls balance between maximizing margin & minimizing training error
* Small -> wide „street“ seperating 2 classes but datapoints might be within boundary -> larger training error but better generalization
* Large -> small „street“ -> lower trianing error but lower generalization
 Gamma = influences how decision boundary adapts to different datapoints = defines how many dp considered for hyperplane
* Small -> considers many dp
* Large -> conisders few dp

Question 31

What where your optimal hyperparameters? What did they tell you? SVM

Answer 31

 C = 100
* Relatively high -> small street, lower training error, but lower generalization
 gamma = 0.0001
* relatively low -> decision boundary less adjusted to individual datapoints, considers a broader range of training instances in determining the decision boundary -> considers more instances -> more generalizable

Question 32

Accuracy

Answer 32

o Accuracy: measures overall correctones of model, ratio of correctly predicted instances to total number of instances

Question 33

Precision

Answer 33

o Precision: measures model ability to predict positive instances correctly -> presents ratio of false positives

Question 34

Recall

Answer 34

o Recall: ratio of true positives to the sum of true positives and false negatives > presents ratio of false negative

Question 35

Which CNN models did you choose? And what are the differences?

Answer 35

CNN1: self-developed model <-> CNN2: adapted from similar paper

Question 36

Differences between CNN models

Answer 36

o CNN1:
 padding (because of cropping)
 more layers (5 conv., 5 maxpooling, 2 dropout, 2 dense)
 dropout
 smaller kernel size (for malaria important -> consider local features, don’t miss info)
 max pooling (<-> average) (better because max focuses on most important / striking features in surrounding)
 2 (<-> 3) dense layers
 Dense layers with 512 units (<-> 256), more information -> capture more complex patterns & nuances in the input (also due to larger image size)
 dense layers have activation function (introduction of non-linearity)

Question 37

Potential explanation based on Differences from CNN2 paper:

Answer 37

o Other parasite
o different images (background etc.)
o no cropping
o smaller images (44x44, we have 128 x 128 -> can have more pooling -> using their architecture will not work (still large images with few pooling layers)

Question 38

Which kernel size should you use?

Answer 38

a. In our case: used small one (3,3) -> because malaria part can be small
b. Also don’t want to risk to miss some information, pattern that indicates malaria

Question 39

Why always relu?

Answer 39

o introduce non-linearity into the network, allowing it to learn complex patterns and gradients
o computationally efficient
o avoiding vanishing gradients
o Images typically consist of pixel intensities ranging from 0 to 255 -> black = 0, dont need negative values

o Non-linearity: For negative inputs (x < 0) -> output 0. This non-linear behavior introduces an element of discontinuity, breaking the linearity of the network
o For positive: still linear -> can capture both non-linearity & linearity

Question 40

What is padding? Why used? -

Answer 40

0 boarders around image pixels, info on boarder, we had boundary boxes -> wanted to classify well

Question 41

Why did CNN 2 use average and max pooling and which one is better

Answer 41

• Max pooling better for image classification
• features tend to encode the spatial presence of some pattern or concept over the different tiles of the feature map (hence, the term feature map),
• more informative to look at maximal presence of different features than at their average presence
• Full contrast

Question 42

Why not KNN?

Answer 42

• Does not scale well: As dataset grows, KNN becomes increasingly inefficientcompromising overall model performance. (Scaling problem)
• prone to overfitting
• Curse of dimensionality: volume of space grows exponentially with dimensions, Need more points to ‘fill’ a high-dimensional volume

Question 43

Why not Random Forest?

Answer 43

• RF better for multi-class classification
• not so well with high-dimensional data compared to SVM
• research showed that RF performs better if 10 to 100 features, SVM if > 100 features (paper that compares which papers used RF or SVM & how they performed)
• could be a good option as well (but literature said that SVM outperformed for Malaria detection)

Question 44

Logistic regression

Answer 44

• linear model that’s why can’t really work
• risk of overfitting if many independent variable
• limited in capturing complexity

Question 45

In can why use dropout and early stopping?

Answer 45

address different aspects of overfitting
* Early stopping: control the capacity of the model by stopping training before it starts overfitting
* Dropout: introduces randomness during training, preventing the network from relying too heavily on any specific set of features or neurons