AI Reading Group
Every Friday 12:00-12:30 UTC+1
We are a research group that gathers every friday to discuss one research paper in the area of AI.
We are interested in any kind of AI applied to any kind of data (text, images, graphs, video, tabular, you name it!).
NL is All a Graph Needs
Paper: Natural Language is All a Graph Needs
Presenter: Filipe Assunção
Date: 15/09/2023
Acoustic Side Channel Attack
Paper: A Practical Deep Learning-Based Acoustic Side Channel Attack on Keyboards
Presenter: Bartlomiej Matejczyk
Date: 15/09/2023
Bug Finder
Paper: Engineering a Formally Verified Automated Bug Finder
Presenter: Arthur Correnson
Date: 16/06/2023
LIMA
Paper: LIMA: Less Is More for Alignment
Presenter: João Nadkarni
Date: 02/06/2023
CodeGen2
Paper: CodeGen2: Lessons for Training LLMs on Programming and Natural Languages
Presenter: João Lages
Date: 26/05/2023
FOCAL
Paper: Target-Driven Compositional Concolic Testing with Function
Summary Refinement for Effective Bug Detection
Presenter: Alexandre Lemos
Date: 19/05/2023
Colossus
Paper: Deferred concretization in symbolic execution via fuzzing
Presenter: Joana Coutinho
Date: 12/05/2023
DGM
Paper: DGM: A deep learning algorithm for solving partial differential equations
Presenter: Bartlomiej Matejczyk
Date: 05/05/2023
Generative Agents
Paper: Generative Agents: Interactive Simulacra of Human Behavior
Presenter: André Menezes
Date: 21/04/2023
CoLT5
Paper: CoLT5: Faster Long-Range Transformers with Conditional Computation
Presenter: João Nadkarni
Date: 14/04/2023
Planning with LLMs for Code Generation
Paper: Planning with Large Language Models for Code Generation
Presenter: João Lages
Date: 31/03/2023
UniLayout
Paper: UniLayout: Taming Unified Sequence-to-Sequence Transformers for Graphic Layout Generation
Presenter: Vasco Almeida
Date: 24/03/2023
Pretraining Language Models with Human Preferences
Paper: Pretraining Language Models with Human Preferences
Presenter: Samuel Arcadinho
Date: 13/03/2023
Jaint
Paper: Jaint: A Framework for User-Defined Dynamic
Taint-Analyses based on Dynamic Symbolic
Execution of Java Programs
Presenter: Alexandre Lemos
Date: 10/03/2023
Relational Memory Augmented LMs
Paper: Relational Memory Augmented Language Models
Presenter: Bartlomiej Matejczyk
Date: 03/03/2023
KLEE
Paper: KLEE: unassisted and automatic generation of high-coverage tests for complex systems programs
Presenter: Joana Coutinho
Date: 24/02/2023
H3
Paper: Hungry Hungry Hippos: Towards Language Modeling with State Space Models
Presenter: André Menezes
Date: 17/02/2023
JetStream
Abstract: JetStream: Automatic Optimization of Virtual Switch Rulesets
Presenter: Margarida Ferreira
Date: 10/02/2023
REPLUG
Paper: REPLUG: Retrieval-Augmented Black-Box Language Models
Presenter: João Nadkarni
Date: 03/02/2023
Onception
Paper: Onception: Active Learning with Expert Advice for Real World Machine Translation
Presenter: Vania Mendonça
Date: 27/01/2023
RLHF
Paper: Reinforcement Learning from Human Feedback
Presenter: João Lages
Date: 13/01/2023
Graph-MLPMixer
Paper: A Generalization of ViT/MLP-Mixer to Graphs
Presenter: Samuel Arcadinho
Date: 06/01/2023
Faster matrix multiplication with RL
Paper: Discovering faster matrix multiplication algorithms with reinforcement learning
Presenter: Bartlomiej Matejczyk
Date: 23/12/2022
Personalizing and Customizing AI Explanations for Clinicians
Paper: Personalizing and Customizing AI Explanations for Clinicians
Presenter: Francisco Calisto
Date: 16/12/2022
Concolic Execution for WebAssembly
Paper: Concolic Execution for WebAssembly
Presenter: Filipe Marques
Date: 09/12/2022
The Ultimate Optimizer
Paper: Gradient Descent: The Ultimate Optimizer
Presenter: André Menezes
Date: 02/12/2022
ROME
Paper: Locating and Editing Factual Associations in GPT
Presenter: João Nadkarni
Date: 25/11/2022
Arbiter
Paper: Arbiter: Bridging the Static and Dynamic Divide in Vulnerability Discovery on Binary Programs
Presenter: Alexandre Lemos
Date: 18/11/2022
Calibration for Few-Shot
Paper: Calibrate Before Use: Improving Few-Shot Performance of Language Models
Presenter: Catarina Belém
Date: 11/11/2022
GLITCH
Paper: GLITCH: Automated Polyglot Security Smell Detection in Infrastructure as Code
Presenter: Nuno Saavedra
Date: 11/04/2022
LLMs as Prompt Engineers
Paper: Large Language Models are Human-Level Prompt Engineers
Presenter: Jakub Zavrel
Date: 28/10/2022
Understanding HTML with LLMs
Paper: Understanding HTML with Large Language Models
Presenter: David Aparício
Date: 21/10/2022
Minimal Contrastive Editing (MICE)
Paper: Explaining NLP Models via Minimal Contrastive Editing (MICE)
Presenter: João Lages
Date: 14/10/2022
Phenaki
Paper: Phenaki: Variable Length Video Generation from Open Domain Textual Descriptions
Presenter: Samuel Arcadinho
Date: 07/10/2022
egg: Fast and Extensible Equality Saturation
Paper: egg: Fast and Extensible Equality Saturation
Presenter: Miguel Neves
Date: 30/09/2022
Beyond neural scaling laws
Paper: Beyond neural scaling laws: beating power law scaling via data pruning
Presenter: André Menezes
Date: 23/09/2022
Diachronic language models
Paper: Diachronic language models
Presenter: Daniel Loureiro
Date: 16/09/2022
The Web Is Your Oyster
Paper: The Web Is Your Oyster - Knowledge-Intensive NLP against a Very Large Web Corpus
Presenter: Sofia Aparicio
Date: 02/08/2022
Describing textual differences
Paper: Describing Differences between Text Distributions with Natural Language
Presenter: David Aparício
Date: 26/08/2022
ExpertosLF
Paper: ExpertosLF: Dynamic Late Fusion of CBIR Systems using Online Learning with Relevance Feedback
Presenter: Soraia Meneses Alarcão
Date: 12/08/2022
CONFETTI
Paper: CONFETTI: Amplifying Concolic Guidance for Fuzzers
Presenter: Alexandre Lemos
Date: 05/08/2022
SVTS: Scalable Video-to-Speech Synthesis
Paper: SVTS: Scalable Video-to-Speech Synthesis
Presenter: Rodrigo Mira
Date: 29/07/2022
Assessing robustness in multimodal transportation systems
Paper: Assessing robustness in multimodal transportation systems
Presenter: João Aparício
Date: 22/07/2022
SSL ViTs with DINO
Paper: Emerging Properties in Self-Supervised Vision Transformers
Presenter: Lourenço Vaz Pato
Date: 15/07/2022
Prompt Adapters
Paper: P-Adapters: Robustly Extracting Factual Information from Language Models with Diverse Prompts
Presenter: João Lages
Date: 08/07/2022
Imagen
Paper: Photorealistic Text-to-Image Diffusion Models with Deep Language Understanding
Presenter: João Nadkarni
Date: 01/07/2022
CoNT
Paper: CoNT: Contrastive Neural Text Generation
Presenter: Samuel Arcadinho
Date: 24/06/2022
Multilingual Clustering
Paper: Simplifying Multilingual News Clustering Through Projection From a Shared Space
Presenter: João Santos
Date: 03/06/2022
BLITZ
Paper: BLITZ: Compositional Bounded Model Checking for Real-World Programs
Presenter: Miguel Neves
Date: 27/05/2022
SimGRACE
Paper: SimGRACE: A Simple Framework for Graph Contrastive Learning without Data Augmentation
Presenter: André Menezes
Date: 20/05/2022
Generating Visual Explanations
Paper: Generating Visual Explanations
Presenter: Isabel Rio-Torto
Date: 13/05/2022
Chain of Thought Prompting
Paper: Chain of Thought Prompting Elicits Reasoning in Large Language Models
Presenter: Mariana Lourenço
Date: 06/05/2022
Input Invariants
Paper: Input Invariants
Presenter: Alexandre Lemos
Date: 29/04/2022
Learning to Scaffold
Paper: Learning to Scaffold: Optimizing Model Explanations for Teaching
Presenter: Patrick Fernandes & Marcos Treviso
Date: 22/04/2022
EvoKG
Paper: EvoKG: Jointly Modeling Event Time and Network Structure for Reasoning over Temporal Knowledge Graphs
Presenter: David Aparício
Date: 08/04/2022
LayoutGMN
Paper: LayoutGMN: Neural Graph Matching for Structural Layout Similarity
Presenter: Filipe Assunção
Date: 01/04/2022
SpeechPainter
Paper: SpeechPainter: Text-conditioned Speech Inpainting
Presenter: Sofia Aparício
Date: 25/03/2022
Once-for-All (OFA)
Paper: Once-for-All: Train One Network and Specialize it for Efficient Deployment
Presenter: João Lages
Date: 18/03/2022
VICReg
Paper: VICReg: Variance-Invariance-Covariance Regularization for Self-Supervised Learning
Presenter: André Menezes
Date: 25/02/2022
XAI in Black Boxes
Paper: Stop Explaining Black Box Machine Learning Models for High Stakes Decisions and Use Interpretable Models Instead
Presenter: Mariana Lourenço
Date: 18/02/2022
XAI with MaxSAT
Paper: Using MaxSAT for Efficient Explanations of Tree Ensembles
Presenter: Alexandre Lemos
Date: 11/02/2022
InstructGPT
Paper: Training Language Models to Follow Instructions with Human Feedback
Presenter: David Aparício
Date: 04/02/2022
Grokking
Paper: Grokking: Generalization Beyond Overfitting on Small Algorithmic Datasets
Presenter: João Lages
Date: 28/01/2022
RETRO
Paper: Improving language models by retrieving from trillions of tokens
Presenter: João Nadkarni
Date: 21/01/2022
AL @ Feedzai
Paper: Active learning for imbalanced data under cold start
Presenter: Ricardo Barata
Date: 07/01/2022
VATT
Paper: VATT: Transformers for Multimodal Self-Supervised Learning from Raw Video, Audio and Text
Presenter: Filipe Assunção
Date: 17/12/2021
MAUVE
Paper: MAUVE: Measuring the Gap Between Neural Text and Human Text using Divergence Frontiers
Presenter: Samuel Arcadinho
Date: 10/12/2021
CORGI
Paper: CORGI: Content-Rich Graph Neural Networks with Attention
Presenter: André Menezes
Date: 26/11/2021
NL Understanding
Paper: Climbing towards NLU: On Meaning, Form, and Understanding in the Age of Data
Presenter: Mariana Lourenço
Date: 12/11/2021
LEGO
Paper: LEGO: Latent Execution-Guided Reasoning for Multi-Hop Question Answering on Knowledge Graphs
Presenter: David Aparício
Date: 05/11/2021
Fuzzing Symbolic Expressions
Paper: Fuzzing Symbolic Expressions
Presenter: Alexandre Lemos
Date: 29/10/2021
TimeGAN
Paper: Time-series Generative Adversarial Networks
Presenter: João Lages
Date: 01/10/2021
CodeT5
Paper: CodeT5: Identifier-aware Unified Pre-trained Encoder-Decoder Models for Code Understanding and Generation
Presenter: João Nadkarni
Date: 24/09/2021
Debuggable DL
Paper: Leveraging Sparse Linear Layers for Debuggable Deep Networks
Presenter: Filipe Assunção
Date: 16/09/2021
Neves' PhD Thesis
Paper: Constraint-Based Multi-Objective Combinatorial Optimization
Presenter: Miguel Neves
Date: 10/09/2021
GUIGAN
Paper: GUIGAN: Learning to Generate GUI Designs Using Generative Adversarial Networks
Presenter: David Aparício
Date: 20/08/2021
Neural Architecture Search (NAS)
Paper: Neural Architecture Search without Training
Presenter: Filipe Assunção
Date: 13/08/2021
Effects of duplicated data
Paper: Deduplicating Training Data Makes Language Models Better
Presenter: João Lages
Date: 06/08/2021
Scout vs GRIDS
Paper1: Scout: Rapid Exploration of Interface Layout Alternatives through High-Level Design Constraints
Paper2: GRIDS: Interactive Layout Design with Integer Programming
Presenter: Alexandre Lemos
Date: 30/07/2021
UniT
Paper: UniT: Multimodal Multitask Learning with a Unified Transformer
Presenter: Sofia Aparício
Date: 16/07/2021
SimSiam
Paper: Exploring Simple Siamese Representation Learning
Presenter: André Menezes
Date: 09/07/2021
CodeBERT
Paper: CodeBERT: A Pre-Trained Model for Programming and Natural Languages
Presenter: João Nadkarni
Date: 02/07/2021
Dynamic Software Updates
Paper: Synthesizing Object State Transformers for Dynamic Software Updates
Presenter: David Aparício
Date: 18/06/2021
Inductive Program Synthesis
Paper: Knowledge Refactoring for Inductive Program Synthesis
Presenter: Alexandre Lemos
Date: 07/06/2021
BYOL
Paper: Bootstrap Your Own Latent: A New Approach to Self-Supervised Learning
Presenter: João Lages
Date: 28/05/2021
Pre-Training GNNs
Paper: Strategies for Pre-training Graph Neural Networks
Presenter: Filipe Assunção
Date: 13/05/2021
SSL of GNNs
Paper: Self-Supervised Learning of Graph Neural Networks: A Unified Review
Presenter: André Menezes
Date: 23/04/2021
Video Compression
Paper: End-to-End Learning of Video Compression Using Spatio-Temporal Autoencoders
Presenter: Jorge Pessoa
Date: 09/04/2021
BluePencil
Paper: On the Fly Synthesis of Edit Suggestions
Presenter: Miguel Neves
Date: 26/03/2021
Transformers for Time-Series
Paper: Temporal Fusion Transformers for Interpretable Multi-horizon Time Series Forecasting
Presenter: João Nadkarni
Date: 12/03/2021
MTFuzz
Paper: MTFuzz: fuzzing with a multi-task neural network
Presenter: David Aparício
Date: 12/03/2021
RAT-SQL
Paper: RAT-SQL: Relation-Aware Schema Encoding and Linking for Text-to-SQL Parsers
Presenter: João Lages
Date: 05/03/2021
BUSTLE
Paper: BUSTLE: Bottom-up program-Synthesis Through Learning-guided Exploration
Presenter: José Amaral
Date: 26/02/2021
GANs through t-SNE
Paper: Demonstrating the Evolution of GANs through t-SNE
Presenter: Victor Costa
Date: 19/02/2021
ADnEV
Paper: ADnEV: Cross-Domain Schema Matching using Deep Similarity Matrix Adjustment and Evaluation
Presenter: Filipe Assunção
Date: 12/02/2021
ALICE-ILP Code Search
Paper: Active Inductive Logic Programming for Code Search
Presenter: David Aparício
Date: 05/02/2021
FL at Scale
Paper: Towards federated Learning at scale: system design
Presenter: Filipe Assunção
Date: 29/01/2021
Peregrine
Paper: Peregrine: a pattern-aware graph mining system
Presenter: Miguel Neves
Date: 22/01/2021