As companies are becoming more and more data-driven with more focus on building machine learning teams and creating models, it’s important to understand the challenges that ML projects present in order to properly address them with good standardized practices. The process of building a machine learning solution is complex and involves many steps that you may be familiar with: Collecting and processing raw data, Analysing the data, Processing the data for training, constructing, train, and test the model, detect bias and analyze performance to validate the model, and finally deploy and monitor the model. It’s clear how work-extensive the entire process can get when you need to repeat it multiple times as business needs and data are constantly changing. This is where MLOps comes in.
In the history of engineering and machine learning, choosing transparent models that are interpretable for humans or end-users is essential. Practically, it means using transparent data sources and simple and easy to interpret models like linear models and decision trees or even rule-based systems despite of their limitations due to real-world scenarios where observations are nonlinear and very specific. With the massive growth of machine learning and deep learning popularity, models complexity, and the spread of AI in all fields, it has became crucial to have approaches and mechanisms to explain models and interpret accurate and inaccurate predictions.
In the era of digitalization, Most companies have various sources of customers feedback, social media, call logs, mobile apps, to name a few. Therefore, analyzing such feedback to come up with actionable insights, is becoming essential for any business with an online presence.
Bayesian Optimization is a useful tool for optimizing an objective function thus helping tuning machine learning models and simulations. Instead of using standard approaches like random search or grid search which are usually expensive or slow to do and where the objective function is a black box (we can not analytically express f or know its derivative), Bayesian Optimization comes in hand to efficiently trades of between exploration and exploitation to find a global optimum is a minimum of number of steps. It mainly rely on the idea of Bayes theorem , posterior = likelihood * prior, to quantify the beliefs about an unknown objective function given samples from the domain \(D\) and their evaluation via the objective function \(f\). Bayesian optimization incorporates prior belief about \(f\) and updates the prior with samples drawn from f to get a posterior that better approximates \(f\). (\(P(f|D) = P(D|f) * P(f)\)).
Human Activity Recognition (HAR) plays an important role in real life applications primarily dealing with human-centric problems like healthcare and eldercare. It has seen a tremendous growth in the last decade playing a major role in the field of pervasive computing. In the recent years, lot of data mining techniques has evolved in analyzing the huge amount of data related to human activity, more specifically, machine learning methods have been been previously employed for recognition include Naive Bayes, SVMs, Threshold-based, Markov chains and deep learning models. Indeed, One important part of the prediction is the selection of suitable models, but also a good selection of relevant features would be benificial in the process of building an accurate model.
Music genre classification has been an interesting problem in the field of Music Information Retrieval (MIR). In this tutorial, We will try to classify music genre using hidden Markov models which are very good at modeling time series data. As Music audio files are time series signals, we expect that HMMs will suit our needs and give us an accurate classification. An HMM is a model that represents probability distributions over sequences of observations. We assume that the outputs are generated by hidden states. To learn more about HMM click click here. Also, you can find a good explanation for HHMs in the time series use case here.
In this tutorial, we will design and implement a deep learning model that learns to recognize digits from sign language. We will be using a convolution neural network built using Keras API and trained on the Turkey Ankara Ayrancı Anadolu High School’s Sign Language Digits Dataset. Many thanks to Arda Mavi for sharing this dataset.