Various evaluation metrics are used for evaluating the effectiveness of a recommender. We will focus mostly on ranking related metrics covering HR (hit ratio), MRR (Mean Reciprocal Rank), MAP (Mean Average Precision), NDCG (Normalized Discounted Cumulative Gain). — Recommender systems eventually output a ranking list of items regardless of different modelling choices. So it is important to look at how to evaluate directly ranking quality instead of other proxy metrics like mean squared error, etc.

A small tutorial or introduction about common loss functions used in machine learning, including cross entropy loss, L1 loss, L2 loss and hinge loss. Practical details are included for PyTorch. — Cross Entropy Cross entropy loss is commonly used in classification tasks both in traditional ML and deep learning. Note: logit here is used to refer to the unnormalized output of a NN, as in Google ML glossary. However, admittedly, this term is overloaded, as discussed in this post.

A gentle intro into Matrix Factorization techniques in Recommender Systems, including FunkSVD , SVD++, and Non-negative Matrix Factorization — Introduction RS is to match items to users. The starting point is a user-item matrix filled with values representing either explicit feedback (user provided ratings) or implicit feedback (count of clicks, number of visits, watch time, etc.). One can also pose this question as a matrix filling problem: given the known…

We go back to the basics and look at Nearest Neighbour methods in the context of density estimation and recommender systems — Introduction Given the recent hyped progress in deep learning based approaches in AI research especially in NLP and CV, it is time to go back to the fundamentals. Many of the fundamental methods, such as neighbourhood based methods are surprisingly effective in many areas, even today. We take a brief look…

ML Pipeline is an important feature provided by Scikit-Learn and Spark MLlib. It unifies data preprocessing, feature engineering and ML model under the same framework. This abstraction drastically improves maintainability of any ML project, and should be considered if you are serious about putting your model in production. — Let’s start by outlining some main benefits of using ML pipeline, already.

Hashing maps objects into different bins. Unlike conventional hashing functions which minimize collision probability, locality sensitive hashing functions maximize it for similar objects. In other words, for a given distance measure, similar items are more likely to be mapped to the same bin with LSH. This way, we can find…

Bloom filter is a probabilistic data structure designed to tell you if a member is in a set in a highly memory and time efficient manner. In that sense, it is similar to a set, but it does it with an adjustable false positive rate. In bloom filter, false positives…

Deep work is valuable, rare and even meaningful in a digital world full of noise and distraction. Now, quit social media and embrace boredom. — Introduction Yuval Noah Harari, the author of Sapiens (an international bestselling phenomenon), has no smartphone. However, that doesn’t stop him from making bold and insightful and somewhat provocative predictions about our very future of humanity and how AI and bio-technologies could join hands and shape history. …

Hey DeepLens fans! In case you forgot your SSH password initially set up up for your DeepLens like I did, sad face. Here are a few steps that I compiled from different resources, proven to be working. Items to prepare before you proceed: usb keyboard micro hdmi to hdmi cable monitor with hdmi input Steps to follow: press ESC key until bios screen comes up select continue and hit enter press ESC once (important step!), then select the SECOND option with “recovery mode” at the end