Constrained Monotonic Neural Networks & Introduction to FastKafka - airt

This time we have two topics, presented by airt:

• Constrained Monotonic Neural Networks
• Introduction to FastKafka

Constrained Monotonic Neural Networks are a new approach to building monotonic neural networks by airt team. Their paper got accepted at ICML 2023 conference.

FastFafka is a Python library for easy connection to the Kafka stream platform. It can be used for putting ML models in production.

1. Constrained Monotonic Neural Networks

Description
Deep neural networks are becoming increasingly popular in approximating arbitrary functions from noisy data. But wider adoption is being hindered by the need to explain such models and to impose additional constraints on them. Monotonicity constraint is one of the most requested properties in real-world scenarios and is the focus of this paper. One of the oldest ways to construct a monotonic fully connected neural network is to constrain its weights to be non-negative while employing a monotonic activation function. Unfortunately, this construction does not work with popular non-saturated activation functions such as ReLU, ELU, SELU, etc, as it can only approximate convex functions. We show this shortcoming can be fixed by employing the original activation function for a part of the neurons in the layer and employing its point reflection for the other part. Our experiments show this approach of building monotonic deep neural networks has matching or better accuracy when compared to other state-of-the-art methods such as deep lattice networks or monotonic networks obtained by heuristic regularization. This method is the simplest one in the sense of having the least number of parameters, not requiring any modifications to the learning procedure or steps post-learning steps.
The academic paper is accepted at ICML 2023, the preprint is available at:
https://arxiv.org/abs/2205.11775

Presenter
Davor Runje is a seasoned software engineer, computer scientist, and serial entrepreneur. Most recently, he co-founded airt, an Artificial Intelligence platform as a service specialized for event data - one of the most common formats in business applications. Prior to that, he co-founded DRAP in 2008, one of Croatia's most awarded digital agencies, acquired in 2018 by Imago Ogilvy. In 1997, he co-founded PlayMedia Systems, the leading provider of AMP MP3 playback technology on the global market. He invented and led the development of several DRM-based and DRM-free systems for digital audio distribution based on AMP® MP3 decoding engine for clients including AOL, STMicroelectronics, and Napster. As a PhD student working with Dean Rosenzweig and Yuri Gurevich at Microsoft Research, he applied the interactive Abstract State Machine thesis to multiprocessor/multicore programming. He’s the author of 20 publications in theoretical computer science, two granted and several pending international patents.

2. Introducing FastKafka

Description
FastKafka is a Python library that makes it easy to connect to Apache Kafka queues and send and receive messages. In this talk, we will introduce the library and its features for working with Kafka queues in Python. We will discuss the motivations for creating the library, how it compares to other Kafka client libraries, and how to use its decorators to define functions for consuming and producing messages. We will also demonstrate how to use these functions to build a simple application that sends and receives messages from the queue. This talk will be of interest to Python developers looking for an easy-to-use solution for working with Kafka.
The documentation of the library can be found here: https://fastkafka.airt.ai/
The talk was initially presented at PyCon DE/PyData Berlin 2023.

Presenter
Tvrtko Sternak is currently working as a Python developer at airt. In the past three years, I have gained valuable experience in the industry, including a year working on a microservice product that uses Apache Kafka for communication between services.
I am a strong believer in the power of open-source software, and I enjoy learning from the open-source community, hopefully also contributing more this year :). My interests in the field at the moment include machine learning, model deployment, Apache Kafka, and advanced Python programming.
In my free time, I enjoy reading fantasy books, staying active through biking and hitting the gym, and watching comedy-drama TV shows.

Join us at the meetup!
When: 10.5.2023 (Wednesday), 18:00
Location: BIRD incubator, Ul. Franje Petračića 4, 10110, Zagreb
Duration: 2 x 30 min presentations + Q&A

Meetup will be held live. Reserve your spot through the meetup platform.
After the presentation, you are invited to hang out with us (drinks and food powered by airt).