Bahram Lavi

Dr. Bahram Lavi


Research Focus & Metrics

Academic Career Length

7Years (2020-Present)

(Calculated once Scilit publication data is synced: current year − year of first publication + 1)

Representative Publications

View All Publications
18 January 2025 in Engineering Applications of Artificial Intelligence
Verified
Journal Article
Closed access
JBJoão Roberto Bertini
João Roberto Bertini
BLBahram Lavi
Bahram Lavi

1 January 2022
Verified
Conference Paper
Open access
BLBahram Lavi
Bahram Lavi
ETEric K. Tokuda
Eric K. Tokuda
FMFelipe Moreno-Vera
Felipe Moreno-Vera
LNLuis Nonato
Luis Nonato
CSClaudio Silva
Claudio Silva
Jorge PocoJorge Poco
Jorge Poco

21 October 2021
Verified
Book Chapter
Closed access
PDF file
FMFelipe Moreno-Vera
Felipe Moreno-Vera
BLBahram Lavi
Bahram Lavi
Jorge PocoJorge Poco
Jorge Poco

2 April 2021 in IEEE Transactions on Information Forensics and Security
Verified
Journal Article
Closed access
CRCaroline Mazini Rodrigues
Caroline Mazini Rodrigues
Aurea Soriano-VargasAurea Soriano-Vargas
Aurea Soriano-Vargas
BLBahram Lavi
Bahram Lavi
ARAnderson Rocha
Anderson Rocha
ZDZanoni Dias
Zanoni Dias
Information coming from social media is vital to the understanding of the dynamics involved in multiple events such as terrorist attacks and natural disasters. With the spread and popularization of cameras and the means to share content through social networks, an event can be followed through many different lenses and vantage points. However, social media data present numerous challenges, and frequently it is necessary a great deal of data cleaning and filtering techniques to separate what is related to the depicted event from contents otherwise useless. In a previous effort of ours, we decomposed events into representative components aiming at describing vital details of an event to characterize its defining moments. However, the lack of minimal supervision to guide the combination of representative components somehow limited the performance of the method. In this paper, we extend upon our prior work and present a learning-from-data method for dynamically learning the contribution of different components for a more effective event representation. The method relies upon just a few training samples (few-shot learning), which can be easily provided by an investigator. The obtained results on real-world datasets show the effectiveness of the proposed ideas.

11 December 2025 in IEEE Access
Verified
Journal Article
Open access
BLBahram Lavi
Bahram Lavi
JBJoão Roberto Bertini
João Roberto Bertini
LPLuis Oliveira Pires
Luis Oliveira Pires
DSDenis José Schiozer
Denis José Schiozer
Accurate, quick forecasting of petroleum production data in short-term scenarios is a complex challenge that requires the development of reliable predictive models. Traditionally, engineers have relied on numerical simulation techniques that depend on dynamic parameters with physical constraints to estimate reservoir production. However, these methods are time-consuming and cumbersome, especially when modeling multiple scenarios. In contrast, Machine Learning (ML) models, including both shallow and deep approaches, offer a faster alternative, enabling engineers to make informed decisions and optimize production strategies more efficiently. This paper compares the most prominent ML methods in the literature to identify the most accurate approach for short-term forecasting of oil rate (OR), water rate (WR), and gas rate (GR). We evaluated six shallow ML algorithms (Gradient Tree Boosting (GTB), K-Nearest Neighbors (KNN), Elastic Net (ENET), Kernel Ridge Regression (KRR), Multi-Layer Perceptron (MLP), Extreme Gradient Boosting (XGBoost)) and seven deep learning regression algorithms (Long Short-term Memory (LSTM), LSTM+Dense layer, (LSTM+Dense), Convolutional Neural Network (CNN), CNN+LSTM, Bidirectional LSTM (BiLSTM), Bidirectional LSTM+Dense layer (BiLSTM+Dense), and Temporal Convolutional Network (TCN)). We compared these models using a presalt offshore benchmark dataset containing production history from six producer wells. Our work provides a comprehensive comparison of both shallow and deep learning methods, and introduces a rigorous statistical analysis to determine the significance of performance differences, offering a more reliable benchmark than simple empirical observation. Model performance was evaluated using several error metrics and regression plots, and the results were analyzed from both empirical and statistical perspectives. Empirical analysis indicated that the models effectively captured the main signal trends and consistently produced accurate forecasts during the transition from the history to the forecast period. Statistical analysis, conducted using production data and the size of the prediction window, identified LSTM, BiLSTM, and BiLSTM + Dense as the top-performing models for short-term production forecasts. Finally, the model’s generalization capability was rigorously tested using cross-well validation to ensure it accurately captures the underlying field dynamics.

30 April 2020 in arXiv
Verified
Preprint
Open access
PDF file
BLBahram Lavi
Bahram Lavi
IUIhsan Ullah
Ihsan Ullah
MFMehdi Fatan
Mehdi Fatan
ARAnderson Rocha
Anderson Rocha
Intelligent video-surveillance (IVS) is currently an active research field in computer vision and machine learning and provides useful tools for surveillance operators and forensic video investigators. Person re-identification (PReID) is one of the most critical problems in IVS, and it consists of recognizing whether or not an individual has already been observed over a camera in a network. Solutions to PReID have myriad applications including retrieval of video-sequences showing an individual of interest or even pedestrian tracking over multiple camera views. Different techniques have been proposed to increase the performance of PReID in the literature, and more recently researchers utilized deep neural networks (DNNs) given their compelling performance on similar vision problems and fast execution at test time. Given the importance and wide range of applications of re-identification solutions, our objective herein is to discuss the work carried out in the area and come up with a survey of state-of-the-art DNN models being used for this task. We present descriptions of each model along with their evaluation on a set of benchmark datasets. Finally, we show a detailed comparison among these models, which are followed by some discussions on their limitations that can work as guidelines for future research.