Robustness

Currently, we are mainly focused on the generalization properties of deep learning models that encounter distribution shifts in data, but also in datasets exhibiting subpopulations or out-of-distribution environments. In this scenario, we focus in particular on the bias analysis and on deep anomaly detection task.

Antonio Barbalau, Elena Burceanu, Cristian Paduraru, Radu Filipescu, Marius Dragoi, Stefan Smeu, Florin Brad, Andrei Manolache

Message-passing graph neural networks

Message-passing graph neural networks (MPNNs) have emerged as a powerful paradigm for graph-based machine learning. Despite their effectiveness, MPNNs face challenges such as under-reaching and over-squashing, where limited receptive fields and structural bottlenecks hinder information flow in the graph. While graph transformers hold promise in addressing these issues, their scalability is limited due to quadratic complexity regarding the number of nodes, rendering them impractical for larger graphs.

Computer Vision

We tackle fundamental unsupervised learning approaches that we consider to be the key to true unconstrained learning, which best simulates how humans discover the surrounding world. We aim to combine the unsupervised visual perception with supervised cognitive video representation. We want to build systems that understand our world by only watching videos. And we go further, teaching the system to describe, in natural language, the discovered elements.

Elena Burceanu, Emanuela Haller, Elisabeta Oneata, Dragos Tantaru

DeepFake Detection

Deepfake detection plays a crucial role in addressing the social impact of synthetic media manipulation. With the rise of deepfake technology, there is a growing concern about its potential to spread misinformation and deceive the public. Detection efforts are essential to protect individuals and society from the harmful consequences of manipulated content. By using advanced machine learning techniques to uncover inconsistencies and anomalies within deepfake videos, audio recordings, or images, these detection methods act as a shield against the erosion of trust in digital media.

Unsupervised learning of objects from video sequences

We address an essential problem in computer vision, that of unsupervised foreground object segmentation in video, where a main object of interest in a video sequence should be automatically separated from its background. Video object segmentation and object discovery are strongly related tasks, but we tackle the problem from a fully unsupervised perspective, building object representations from raw video sequences. An efficient solution to this task would enable large-scale video interpretation at a high semantic level in the absence of the costly manual labeling.

Graph methods for video processing

Given a video we want to have a good understanding of the scene and to be able to identify the key events in order to extract useful information. Our goal is to capture the complex interactions between multiple entities in a scene. We improve the clasical convolutional models by proposing a graph model, that has a strong, explicit bias towards modeling relationships and at the same time being able to model long range interactions. We also design a novel method, devised at the object-level, that can discover salient regions and we quantitatively show that they correlate with objects locations and help the relational procesing

Multi-task graph

Existing methods in multi-task graphs rely on expensive manual supervision. In contrast, our proposed solution, with consensus shift learning, relies only on pseudo-labels provided by expert models. In our graph, every node represents a task, and every edge learns to transform one input node into another. Once initialized, the graph learns by itself on virtually any novel target domain. An adaptive selection mechanism finds consensus among multiple paths reaching a given node and establishes the pseudo-ground truth at that node. Such pseudo-labels, given by ensemble pathways in the graph, are used during the next learning iteration when single edges distill this distributed knowledge.

Video Captioning

Video captioning is the task of describing a video in natural language. It lies at the intersection of computer vision, natural language processing and machine learning requiring both high level visual comprehension and the ability to produce meaningful sentences. Our goal is to detect objects and events in a video and be capable of understanding the interactions between them in spatial and temporal dimensions.

Natural Language Processing

A large amount of today's data is stored in databases. Building AI tools that facilitate the access to knowledge requires processing of natural language and structured data. We focus on neural approaches for natural language interfaces to databases, in particular structure-aware and semi-supervised methods.

Florin Brad, Andrei Manolache, Ioana Pintilie, Dragos Tantaru, Marius Dragoi

Large Language Models for Assembly

Features learned by pretrained models are crucial for downstream tasks such as anomaly detection, search or classification. Most LLMs (either general purpose or code-based) have little assembly data in their pretraining corpus relative to the entire corpus. Finetuning LLMs on more assembly data should then result in better features. To test this, we developed several downstream classification tasks, where the input is assembly code. Classifiers trained on features from the asmLLM should then outperform classifiers trained on features from base LLMs.

Anomaly detection in Text

Recent deep methods for detecting anomalies in images learn better features of normality in an end-to-end self-supervised setting. These methods train a model to discriminate between different transformations applied to visual data and then use the output to compute an anomaly score. We use this approach for Anomaly Detection in text, by introducing a novel pretext task on text sequences. We learn our model end-to-end, enforcing two independent and complementary self-supervision signals, one at the token-level and one at the sequence-level.

Domain Adaptation for Authorship Verification

The task of identifying the author of a text spans several decades and was tackled using linguistics, statistics, and, more recently, machine learning. We tackle it inspired by the impressive performance gains across a broad range of natural language processing tasks and by the recent availability of the PAN large-scale authorship dataset.

Visual Question Answering

The VQA task refers to answering questions about an image. Existing methods fuse image and text representations and are able to exploit superficial correlations and produce the correct answer, but often for the wrong reason. The recently introduced Visual Commonsense Reasoning dataset facilitates cognition-level understanding on top of recognition, by requiring not only the correct answer, but also a justification for it. We improve existing methods by leveraging structured representations of images.

Natural Language Interface to Databases

Natural Language Interface to Databases (NLIDB) bridges the gap between technical and non-technical users by allowing the latter to query large amounts of structured data through the use of instructions written in natural language.

Despite long-standing research efforts, progress has been slow and widespread adoption has failed to pick up. Data-driven approaches have been hindered by the lack of large parallel corpora to train the models on, but recent datasets alleviate this problem. We seek to improve existing SEQ2SEQ models by leveraging syntax to guide the generation process and by using semi-supervised techniques to overcome the low parallel data regime.

Reinforcement Learning

Within the field of artificial intelligence reinforcement learning seems a natural setting for training agents that interact with the world we are living in. We engage in furthering the field by developing agents able to learn continuously in different environments. We are also investigating models sitting at the intersection of generative models and reinforcement learning.

Florin Gogianu

Learning Representations for Deep Reinforcement Learning

We are interested in furthering the research on learning good representations for Deep Reinforcement Learning towards the goals of more stable optimisation, improvede generalization and better sample complexity.

Malmo AI Challenge

The Microsoft Malmo AI Challenge proposed a time-extended stag hunt scneario build on top of the well-known Minecraft platform. In order to maximize its payoff in such a game an agent needs to predict the level of commitment to a collaborative strategy of the other player, decide on a specific plan and navigate through the environment to execute it.

Our submission ranked second for the AI Summer School placement prize and third for the Microsoft Azure for Research Grant prize.

Cryptography

Lattice-based cryptography is a great promise for post-quantum cryptography. We build advanced primitives whose security relies on the hardness of lattice problems.

Miruna Rosca, Mădălina Bolboceanu, Dacian Stroia

Hardness of lattice problems

Lattice-based cryptography is a great promise for post-quantum cryptography. It aims at harnessing the security of cryptographic primitives in the conjectured hardness of well-identified and well-studied algorithmic problems involving euclidean lattices. In order to build post-quantum cryptographic primitives based on lattices, we actually make use of some intermediate, more versatile problems, Learning With Errors (LWE) and Shortest Integer Solutions (SIS), which are provably at least as hard as classical lattice problems.

To obtain more efficient primitives, different structured variants of LWE and SIS have been introduced. We are interested in studying the hardness of all these problems and giving reductions between them.

Machine Learning team

elena_burceanu.webp

Elena Burceanu

I recently defended my PhD thesis in Computer Vision, on `Efficiently Exploiting Space-Time Consensus for Object Segmentation and Tracking in Video`, at University of Bucharest, under Marius Leordeanu and GheorgheStefanescu supervision. My current research interest is in Generalization problems under Distribution Shifts, with a focus on the anomaly detection task.

florin_brad.webp

Florin Brad

I am interested in neural generative models for natural language processing, especially for code generation. In particular, I am interested in leveraging discrete structure (syntax trees) to improve the expresiveness of the latent space and to guide the generation process.

andrei_manolache.webp

Andrei Manolache

I am a Ph.D. student at the IMPRS-IS and ELLIS, under the co-supervision of Mathias Niepert (University of Stuttgart) and Karsten Borgwardt (ETH Zürich & MPI-B) and a Researcher at Bitdefender’s Theoretical Research Laboratory. My focus is on robustness and trustworthiness of Geometric Deep Learning methods.

marius_dragoi.webp

Marius Drăgoi

I studied at Politehnica University of Bucharest, where I obtained a Bachelor's degree in Computer Science and a Master's degree in Artificial Intelligence. I am currently researching the problem of out-of-distribution sample detection and distributional shift in log data.

florin_gogianu.webp

Florin Gogianu

Currently pursuing a PhD in Reinforcement Learning unde the supervision of Prof. Lucian Bușoniu following an MSc in Artificial Intelligence from University Politehnica of Bucharest and a BSc in Philosophy. I have a broad interest in Reinforcement Learning topics and I am currently focusing on questions regarding sample efficiency in the context of model-free value-based methods with neural network estimators.

stefan_smeu.webp

Stefan Smeu

I studied Mathematics and Computer Science at University of Bucharest, where I got my bachelor's degree and currently pursuing it's Artificial Intelligence Master degree program. I am interested in the unsupervised representation learning branch of deeplearning. Nowadays I research the problem of distributional shift and anomaly detection, mainly on visual data.

elisabeta_oneata.webp

Elisabeta Oneata

I have obtained my PhD in computer vision from the School of Advanced Studies of the Romanian Academy (SCOSAAR), with the thesis “Perceptual Models for Computational Visual Analysis”, under the supervision of Prof. Dr. Cristian Sminchisescu. My work focused on 3d human pose estimation and perception in monocular images and videos. Currently I am interested in automatic detection of deepfake visual content.

ioana_pintilie.webp

Ioana Pintilie

I am an Artificial Intelligence master’s student at the University of Bucharest, where I also obtained my bachelor’s degree in Computer Science. I am interested in anomaly detection for multivariate time series data, diffusion models, and I have a general interest in NLP topics. Currently, my focus is on large language models.

dragos_tantaru.webp

Dragos Tantaru

I am a researcher in the field of Deep Learning, with expertise in Deepfake detection and Large Language Models (LLMs) finetuning and optimization. I'm also very interested in works pertaining to theoretical aspects of Deep Learning. I'm currently pursuing a Master's degree in Artificial Intelligence at the University of Bucharest.

cristi_paduraru.png

Cristian Paduraru

I earned my Bachelor's degree in Computer Science from the University of Bucharest, where I am currently pursuing a Master's program in Artificial Intelligence. My interests include bias identification and correction, as well as model interpretability.

radu_filipescu.jpg

Radu Filipescu

Starting the Artificial Intelligence master program at the University of Bucharest, where I also graduated the Faculty of Mathematics and Computer Science. I am interested in multiple applications of Machine Learning, and now working mainly on Domain Generalization and Distribution Shift problems.

antonio_barbalau.png

Antonio Barbalau

Currently pursuing a PhD focused on weakly-supervised and self-supervised learning at the University of Bucharest. I've been mainly involved in model stealing, video anomaly detection and speech processing. Presently, I'm working on domain generalization.

andu_boldisor.jpg

Dragos (Andu) Boldisor

I am currently studying for my Master's Degree in Artificial Intelligence at University Politehnica of Bucharest. I am presently researching approaches to deepfake detection.

collaborators
andrei_nicolicioiu.webp

Andrei Nicolicioiu

PhD Student @ University of Montreal

teo_poncu.jpg

Teo Poncu

Research Scientist @ University Politehnica of Bucharest, nVidia DLI University Ambassador

tudor_berariu.webp

Tudor Berariu

PhD Student @ Imperial College London

iulia_duta.webp

Iulia Duta

PhD Student @ University of Cambridge

marius_leordeanu.webp

Marius Leordeanu

Prof @ University Politehnica of Bucharest

radu_titiu.webp

Radu Țițiu

PhD in Cryptography from École Normale Supérieure de Lyon