Nasim Soltani

Northeastern University

Author Statistics

Papers

Citation

H-Index

i-10 index

Research Trends

Author Order

Document Type

Co-Authors

Kaushik Chowdhury

The University of Texas at Austin

Stratis Ioannidis

Northeastern University

Georgios B. Giannakis

University of Minnesota

Debashri Roy

The University of Texas at Arlington

Seung-Jun Kim

University of Maryland, Baltimore County

Dibaloke Chanda

Marquette University

Yanzhi Wang

Northeastern University

Jennifer Dy

Northeastern University

Yanyu Li

Shanghai Jiao Tong University

Deniz Erdoğmuş

Northeastern University

Cooperative Institutions

Northeastern University

Boston University

Universidad del Noreste

Spiral Foundation

University of North Carolina at Charlotte

Signal Processing (United States)

Chinese Academy of Sciences

University of Wisconsin–Madison

National Instruments (United States)

Iran University of Science and Technology

Author Statistics

Papers

Citation

H-Index

i-10 index

Research Field

Online Learning of Sparse Gaussian Conditional Random Fields with Application to Prediction of Energy Consumers Behavior

Nasim Soltani

Many real-world networks in different application domains, are efficiently represented using graphs. In many cases, the network graph topology is unobservable and evolves over time. To infer the graph structure from signal observations, Gaussian conditional random field (GCRF) is deployed. To account for temporal dynamics even in an adversarial setting, the framework of online convex optimization is adopted to develop an efficient online algorithm for estimating the GCRF parameters. The proposed algorithm is then used to track consumers energy consumption dependency network and price responsiveness in the context of smart grid energy management. Numerical tests verify that the proposed approach can track the dynamics of network graph.

Unobservable

10.1109/ssp49050.2021.9513739

Cite

Citations (0)

Context-Aware Self-Supervised Learning of Whole Slide Images

IEEE Transactions on Artificial Intelligence (2024)

Milan Aryal Nasim Soltani

Presenting whole slide images (WSIs) as graph will enable a more efficient and accurate learning framework for cancer diagnosis. Due to the fact that a single WSI consists of billions of pixels and there is a lack of vast annotated datasets required for computational pathology, the problem of learning from WSIs using typical deep learning approaches such as convolutional neural network (CNN) is challenging. Additionally, WSIs down-sampling may lead to the loss of data that is essential for cancer detection. A novel two-stage learning technique is presented in this work. Since context, such as topological features in the tumor surroundings, may hold important information for cancer grading and diagnosis, a graph representation capturing all dependencies among regions in the WSI is very intuitive. Graph convolutional network (GCN) is deployed to include context from the tumor and adjacent tissues, and self-supervised learning is used to enhance training through unlabeled data. More specifically, the entire slide is presented as a graph, where the nodes correspond to the patches from the WSI. The proposed framework is then tested using WSIs from prostate and kidney cancers.

Supervised Learning

10.1109/tai.2024.3365779

Cite

Citations (4)

A New Era in Computational Pathology: A Survey on Foundation and Vision-Language Models

arXiv (Cornell University) (2024)

Dibaloke Chanda Milan Aryal Nasim Soltani Masoud Ganji

Recent advances in deep learning have completely transformed the domain of computational pathology (CPath). More specifically, it has altered the diagnostic workflow of pathologists by integrating foundation models (FMs) and vision-language models (VLMs) in their assessment and decision-making process. The limitations of existing deep learning approaches in CPath can be overcome by FMs through learning a representation space that can be adapted to a wide variety of downstream tasks without explicit supervision. Deploying VLMs allow pathology reports written in natural language be used as rich semantic information sources to improve existing models as well as generate predictions in natural language form. In this survey, a holistic and systematic overview of recent innovations in FMs and VLMs in CPath is presented. Furthermore, the tools, datasets and training schemes for these models are summarized in addition to categorizing them into distinct groups. This extensive survey highlights the current trends in CPath and its possible revolution through the use of FMs and VLMs in the future.

Foundation (evidence)

10.48550/arxiv.2408.14496

Cite

Citations (0)

AirID: Injecting a Custom RF Fingerprint for Enhanced UAV Identification using Deep Learning

GLOBECOM 2022 - 2022 IEEE Global Communications Conference (2020)

Subhramoy Mohanti Nasim Soltani Kunal Sankhe Dheryta Jaisinghani Marco Di Felice

We propose a framework called AirID that identifies friendly/authorized UAVs using RF signals emitted by radios mounted on them through a technique called as RF fingerprinting. Our main contribution is a method of intentionally inserting `signatures' in the transmitted I/Q samples from each UAV, which are detected through a deep convolutional neural network (CNN) at the physical layer, without affecting the ongoing UAV data communication process. Specifically, AirID addresses the challenge of how to overcome the channel-induced perturbations in the transmitted signal that lowers identification accuracy. AirID is implemented using Ettus B200mini Software Defined Radios (SDRs) that serve as both static ground UAV identifiers, as well as mounted on DJI Matrice M100 UAVs to perform the identification collaboratively as an aerial swarm. AirID tackles the well-known problem of low RF fingerprinting accuracy in `train on one day test on another day' conditions as the aerial environment is constantly changing. Results reveal 98% identification accuracy for authorized UAVs, while maintaining a stable communication BER of 10 ^-4 for the evaluated cases.

Identification

Signature (topology)

SIGNAL (programming language)

10.1109/globecom42002.2020.9322561

Cite

Citations (28)

Context-Aware Graph-Based Self-Supervised Learning of Whole Slide Images

ICASSP 2022 - 2022 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP) (2022)

Milan Aryal Nasim Soltani

The gigapixel resolution of a single whole slide image (WSI), and the lack of huge annotated datasets needed for computational pathology, makes cancer diagnosis and grading with WSIs a challenging task. Moreover, downsampling of WSIs might result in loss of information critical for cancer diagnosis. Motivated by the fact that context such as topological structures in the tumor environment may contain critical information in cancer grading and diagnosis, a novel two-stage learning approach is proposed. Self-supervised learning is applied to improve training through unlabled data and graph convolutional network (GCN) is deployed to incorporate context from tumor and surrounding tissues. More specifically, we represent the whole slide as a graph, where nodes are patches from the WSIs. The patches in the graph are represented as feature vectors obtained from pre-training the patches in self-supervised learning. The graph is trained using GCN which accounts for the context of each tissue for the cancer grading and classification. In this work, WSIs for prostrate cancer are validated and the model performance is evaluated based on diagnosis and grading of prostrate cancer and compared with ResNet50 as a traditional convolutional neural network (CNN) and multi-instance learning (MIL) as a leading approach in WSI diagnosis.

Grading (engineering)

Upsampling

10.1109/icassp43922.2022.9747899

Cite

Citations (5)

Neural Network-based OFDM Receiver for Resource Constrained IoT Devices

arXiv (Cornell University) (2022)

Nasim Soltani Hai Cheng Mauro Belgiovine Yanyu Li Haoqing Li

Orthogonal Frequency Division Multiplexing (OFDM)-based waveforms are used for communication links in many current and emerging Internet of Things (IoT) applications, including the latest WiFi standards. For such OFDM-based transceivers, many core physical layer functions related to channel estimation, demapping, and decoding are implemented for specific choices of channel types and modulation schemes, among others. To decouple hard-wired choices from the receiver chain and thereby enhance the flexibility of IoT deployment in many novel scenarios without changing the underlying hardware, we explore a novel, modular Machine Learning (ML)-based receiver chain design. Here, ML blocks replace the individual processing blocks of an OFDM receiver, and we specifically describe this swapping for the legacy channel estimation, symbol demapping, and decoding blocks with Neural Networks (NNs). A unique aspect of this modular design is providing flexible allocation of processing functions to the legacy or ML blocks, allowing them to interchangeably coexist. Furthermore, we study the implementation cost-benefits of the proposed NNs in resource-constrained IoT devices through pruning and quantization, as well as emulation of these compressed NNs within Field Programmable Gate Arrays (FPGAs). Our evaluations demonstrate that the proposed modular NN-based receiver improves bit error rate of the traditional non-ML receiver by averagely 61% and 10% for the simulated and over-the-air datasets, respectively. We further show complexity-performance tradeoffs by presenting computational complexity comparisons between the traditional algorithms and the proposed compressed NNs.

10.48550/arxiv.2205.06159

Cite

Citations (0)

Position-Aware Graph-Based Learning of Whole Slide Images

ICASSP 2022 - 2022 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP) (2023)

Milan Aryal Nasim Soltani

Encoding whole slide images (WSI) as graphs is well motivated as it allows the gigapixel resolution WSI to be represented in its entirety for learning. To this end, WSIs can be divided into small patches representing the nodes of a graph. Then, graph-based learning approaches can be deployed for cancer grading and classification. Graph-based learning methods such as graph neural network (GNN) are based on message passing among neighboring nodes. However, they do not account for positional information for each patch and if two patches are located in topologically isomorphic neighborhoods, their embedding is almost identical. In this work, classification of cancer from WSI is performed with positional embedding and graph attention. The proposed approach is based on spline convolutional neural network (CNN) to encode the positional embedding of the nodes in graph classification. The algorithm is then tested with the WSI dataset for grading prostate cancer and kidney cancer. A comparison of the proposed method with leading approaches in cancer diagnosis and grading verify improved performance.

Graph Embedding

ENCODE

10.1109/icassp49357.2023.10096190

Cite

Citations (0)

Real-Time Load Elasticity Tracking and Pricing for Electric Vehicle Charging

IEEE Transactions on Smart Grid (2014)

Nasim Soltani Seung-Jun Kim Georgios B. Giannakis

While electric vehicles (EVs) are expected to provide environmental and economical benefit, judicious coordination of EV charging is necessary to prevent overloading of the distribution grid. Leveraging the smart grid infrastructure, the utility company can adjust the electricity price intelligently for individual customers to elicit desirable load curves. In this context, this paper addresses the problem of predicting the EV charging behavior of the consumers at different prices, which is a prerequisite for optimal price adjustment. The dependencies on price responsiveness among consumers are captured by a conditional random field (CRF) model. To account for temporal dynamics potentially in a strategic setting, the framework of online convex optimization is adopted to develop an efficient online algorithm for tracking the CRF parameters. The proposed model is then used as an input to a stochastic profit maximization module for real-time price setting. Numerical tests using simulated and semi-real data verify the effectiveness of the proposed approach.

Profit maximization

Dynamic Pricing

Maximization

Demand Response

10.1109/tsg.2014.2363837

Cite

Citations (92)

RF Fingerprinting Unmanned Aerial Vehicles With Non-Standard Transmitter Waveforms

IEEE Transactions on Vehicular Technology (2020)

Nasim Soltani Guillem Reus-Muns Batool Salehi Jennifer Dy Stratis Ioannidis

The universal availability of unmanned aerial vehicles (UAVs) has resulted in many applications where the same make/model can be deployed by multiple parties. Thus, identifying a specific UAV in a given swarm, in a manner that cannot be spoofed by software methods, becomes important. We propose RF fingerprinting for this purpose, where a neural network learns subtle imperfections present in the transmitted waveform. For UAVs, the constant hovering motion raises a key challenge, which remains a fundamental problem in previous works on RF fingerprinting: Since the wireless channel changes constantly, the network trained with a previously collected dataset performs poorly on the test data. The main contribution of this paper is to address this problem by: (i) proposing a multi-classifier scheme with a two-step score-based aggregation method, (ii) using RF data augmentation to increase neural network robustness to hovering-induced variations, and (iii) extending the multi-classifier scheme for detecting a new UAV, not seen earlier during training. Importantly, our approach permits RF fingerprinting on manufacturer-proprietary waveforms that cannot be decoded or altered by the end-user. Results reveal a near two-fold accuracy in UAV classification through our multi-classifier method over the single-classifier case, with an overall accuracy of 95% when tested with data under unseen channel. Our multi-classifier scheme also improves new UAV detection accuracy to a near perfect 99%, up from 68% for a single neural network approach.

Robustness

10.1109/tvt.2020.3042128

Cite

Citations (78)

A different approach to use narrowband super-resolution multiple signal classification algorithm on wideband sources

The Journal of the Acoustical Society of America (2010)

Mohammad Asgari Nasim Soltani Ali Riahi

There are varieties of wideband direction-of-arrival (DOA) estimation algorithms. Their structure comprises a number of narrowband ones, each performs in one frequency in a given bandwidth, and then different responses should be combined in a proper way to yield true DOAs. Hence, wideband algorithms are always complex and so non-real-time. This paper investigates a method to derive a flat response of narrowband multiple signal classification (MUSIC) [R. O. Schmidt, IEEE Trans. Antennas Propag., 34, 276-280 (1986)] algorithm in the whole frequencies of given band. Therefore, required conditions of applying narrowband algorithm on wideband impinging signals will be given through a concrete analysis. It could be found out that array sensor locations are able to compensate the frequency variations to reach a flat response of DOAs in a specified wideband frequency.

Narrowband

Wideband

Frequency band

SIGNAL (programming language)

10.1121/1.3268609

Cite

Citations (0)