Faried Abu Zaid | TransferLab — appliedAI Institute

Faried Abu Zaid

Faried firmly believes that understanding the mathematical foundations of Artificial Intelligence is not only important for anyone looking to create reliable AI applications, but also that they are fun and interesting when communicated in an intuitive way alongside real world examples. This conviction was shaped by his journey, which led him from researching the mathematical foundations of computer science to consulting in data science and finally to applied AI research.

Faried's work

Practical anomaly detection

An two-day introduction to unsupervised ML techniques for anomaly detection and their strengths and weaknesses in different application areas.

Symmetry Teleportation for Accelerated Optimization

A novel approach, symmetry teleportation, enhances convergence speed in gradient-based optimization by allowing parameters to traverse …

sbi: the simulation-based inference toolkit

sbi is a Python package for Bayesian parameter inference on simulators. It implements state-of-the-art algorithms and comes with …

SoftFlow: Probabilistic Framework for Normalizing Flow on Manifolds

A simple yet effective method for training continuous and discrete normalizing flows on distributions and point clouds where the support …

Flow Matching for Generative Modeling

Flow matching provides a simulation free method for training continuous normalizing flows. Key ingredients are an implicit definition of …

Modeling Dynamical Systems with Geometric Clifford Algebra Networks

The paper investigates the use of Clifford algebras in deep learning. The authors define a general parametric group action layer, which …

Introduction to Explainable AI

Safety and reliability concerns are major obstacles for the adoption of AI in practice. In addition, European regulation will make …

Sampling Free Epistemic Uncertainty Through Approximated Variance Propagation

Noise injection methods such as dropout are popular ways of implicitly capturing epistemic uncertainty in neural networks. Usually, noise …

Optimization Based Modelling

Optimization layers allow incorporating downstream optimization problems into neural network training. The performance gap between the …

Towards Generative Modelling of Multivariate Extremes and Tail Dependence

Modeling heavy-tailed marginal distributions and asymmetric tail dependence with normalizing flows and copula theory.

Our take on IJCAI 2022

The International Joint Conference on Artificial Intelligence (IJCAI) is a premier event for researchers working in all areas of AI.

Neuro Symbolic Verification of Deep Neural Networks

The neuro-symbolic approach to verification of neural networks uses reference networks to represent high level concepts. This allows the …

Probabilistic Inference With Algebra and Logic Constraints

Hybrid probabilistic inference allows performing probabilistic inference with algebra and logic constraints. Recent advances with emphasis …

Complexity of Approved Winners

A thorough complexity theoretic analysis of computing approved winners under a distance metric. Various election schemes are considered, …

Network Creation With Homophilic Agents

Segregation describes the tendency of a heterogeneous group to form homogeneous clusters. To understand this phenomenon better, equilibria …

Robust Solutions for Multi-Defender Stackelberg Security Games

Classical formulations of Multi-defender Stackelberg security games have the unpleasant property that the solutions are highly sensitive to …

Tesselation Filtering ReLu Networks

An analysis of the shape complexity of the decision surface of a ReLU neural network. A framework for decomposing a network along a specific …

Methods and issues in explainable AI

A two-day workshop for ML practitioners wishing to make their models better understandable for themselves and decision makers.

The hidden assumptions and pitfalls of density based anomaly detection

Anomaly detection is a notoriously ill-defined problem. The notion of an anomaly is arguably subjective since it depends to some extent on …

Towards relatable explainable AI

A novel contrastive explanation method for neural network classifiers is proposed and applied in the challenging realm of audio prediction. …

Statistical Rethinking

A beautiful lecture series on Bayesian statistics.

Rethinking Graph Transformers with Spectral Attention

Positional encodings based on the spectrum of the graph laplacian are proposed as a way to generalize global attention transformers to …

Our take on ICLR 2022 - Part 2

ICLR is one of the most important and prestigious conferences in the ML community. It spans all ML topics, from embeddings to Graph neural …

Natural Posterior Networks

A new way to train neural networks using the exponential family for estimates of predictive uncertainty.

Anomaly Transformer: Time Series Anomaly Detection with Association Discrepancy

A new minimax optimization scheme to learn complex and long range associations between the points in a time series. Associations are used …

Escaping limit cycles: Global convergence for constrained nonconvex-nonconcave minimax problems

This paper proposes a new algorithm, curvatureEG+, that converges for a broader range of weak minty variational inequalities.

Graph Augmented Normalizing Flows for Anomaly Detection of Multiple Time Series

A new method for simultaneously detecting anomalies across multiple time series. The structure of a Bayesian network is learned as the …

GreaseLM: Graph REASoning Enhanced Language Models

Capabilities of large language models and massive knowledge graphs are combined through a novel layer-wise interaction between the language …

Variational Methods for Simulation Based Inference

A new framework for likelihood-free inference, i.e. the task of using probabilistic / Bayesian Methods in the absence of tractable …

Evaluation Metrics for Graph Generative Models: Problems, Pitfalls, and Practical Solutions

An investigation of common problems and pitfalls of the popular MMD technique for comparing distributions on graphs. Additionally, the …

Understanding over-squashing and bottlenecks on graphs via curvature

Oversquashing refers to the struggling of graph neural networks with tasks requiring long-distance interactions between nodes. A new notion …

Scalable Bayesian Deep Learning with Modern Laplace Approximations

Training Bayesian neural networks using the Laplacian method for Gaussian posterior approximation.

An introduction to Bayesian methods in ML

A two-day workshop introducing Bayesian modelling, using practical examples and probabilistic programming.

Bayesian ML and Probabilistic Programming 3: Variational Inference

In this talk, we lay the ground for Markov Chain Monte Carlo (MCMC) methods to inferring the posterior distribution, after looking into the …

Isolation Forests: The good, the bad and the ugly

Anomaly detection is one of the main methods behind numerous real life machine learning use cases such as predictive maintenance, network …

Conditional Random Fields with an application to Information Extraction

In this week’s presentation we welcome a new colleague, Faried Abu Zaid, who will discuss Conditional Random Fields (CRFs), a kind of …