Deep Learning and Computer Vision Laura Portell 1 Deep Learning 2 Definition Deep Learning DL is a subset of machine learning which in turn is a subset of artificial intelligence ARTIFICIAL INTELLIGENCE MACHINE LEARNING Deep Learning allows us to train an Artificial Intelligence model to obtain a prediction given a set of inputs DEEP LEARNING 3 Definition DL is inspired by the structure of the human brain This structure is called an artificial neural network Neuronal network take in data train themselves to recognizes the patterns in this data and predict the outputs for a new set of similar data Output Neural Network 4 Neuronal network We will see how a neural network works from an example 9 9 student writes Each digits differently however the human brain knows how to recognize it as the same 9 5 Neuronal network Here is a neural network trained to identify handwritten digits Each number is represented as an image 9 28 pixels s l e x i p 8 2 784 pixels 6 Neuronal network Neurons the core entity of a neural network is where the information processing takes place Each of the 784 pixels gives value to a neuron in the first layer 7 Neuronal network The input layer receives the input data and the Output layer returns the prediction made In this example the output layer represents a digit 8 Neuronal network The hidden layers performed mathematical calculations on our inputs One of the challenges in creating the Neural Network is deciding the number of hidden layers and the number of neurons in each layer 9 Neuronal network The information is transport from one layer to another over connecting channels 10 Neuronal network Each channel has a value attached to it and is called a weighted channel 11 Neuronal network All neurons have a unique number associated with them called a bias 12 Neuronal network This bias is added to the weighted sum of inputs reaching the neuron which is then applied to a function known as the activation function B1 x1 0 3 x2 0 6 Activation Function 13 Neuronal network The result of the activation function determines if the neuron gets activated B1 x1 0 3 x2 0 6 Activation Function 14 Neuronal network Every activated neuron passes on information to the following layers this up till the second last layer B1 x1 0 3 x2 0 6 Activation Function 15 Neuronal network The one neuron activated in the output corresponds to the predict digit The weights and bias are continually adjusted to produce a well trained network 16 Example What Neural Networks See https pair withgoogle com explorables dataset worldviews 17 Explainability 18 19 What is Model Explainability Model explainability refers to the concept of being able to understand the deep learning or machine learning model For example If a healthcare model is predicting whether a patient is suffering from a particular disease or not The medical practitioners need to know what parameters the model is taking into account or if the model contains any bias 20 Why is Model Explainability required Being able to interpret a model increases trust in a machine learning model This becomes all the more important in scenarios involving life and death situations like healthcare law credit lending etc Once we understand a model we can detect if there is any bias present in the model If a healthcare model has been trained on the American population it might not be suitable for Asian people For example 21 Why is Model Explainability required Model Explainability becomes important while debugging a model during the development phase Model Explainability is critical for getting models to vet by regulatory authorities like Food and Drug Administration FDA National Regulatory Authority etc It also helps to determine if the models are suitable to be deployed in real life 22 Computer Vision 23 Definition Computer vision is a field of study which enables computers to replicate the human visual system It s a subset of artificial intelligence which collects information from digital images or videos and processes them to define the attributes The main objective of this branch of artificial intelligence is to teach machines to collect information from pixels 24 How does computer vision work Computer vision works by trying to mimic the human brain s capability of recognising visual information It uses pattern recognition algorithms to train machines on a large amount of visual data The computer then processes input images labels the objects on these images and finds patterns in those objects 25 How does computer vision work Input Sensing Device Interpreting Device Output Computer Vision Human Vision Image mountains sun Image mountains sun Output 26 Input Eye Brain Why is Computer Vision Important From selfies to landscape images we are flooded with all kinds of photos today With the easy connectivity the internet is easily accessible by all today Children are especially susceptible to online abuse and toxicity Apart from automating a lot of functions computer vision also ensures moderation and monitoring of online visual content https blog securly com 2015 11 25 safe kid friendly alterna tives to google and youtube 27 Why is Computer Vision Important Popular search engines like Google and Youtube use computer vision to scan through images and videos to approve them for featuring they have created safe search options in order to protect young kids from inappropriate content on the web e g Google SafeSearch and Youtube Safety Mode 28 Where is Computer Vision used 1 Healthcare Computer vision is widely used in health care Medical diagnosis involves many imaging studies scans and photographs Covid 19 tracking 29 Where is Computer Vision used 2 Automotive Industry Autonomous vehicles are equipped with extensive cameras that film their surroundings over a wide area The then monitored in real time by an image recognition algorithm resulting footage is 30 Where is Computer Vision used 3 Agriculture Modern technologies enable farmers to cultivate ever larger fields efficiently This means that these areas must be checked for pests and plant diseases because if overlooked plant diseases can lead to painful harvest losses and crop failures 31 Where is Computer Vision used 4 Facial Recognition recognition Computer vision also plays an important role in applications of the facial allows technology computers to match images of people s their identities We will see this in detail faces that to 32 Facial Recognition 33 Facial Recognition Facial recognition is a way