Gabor Fodor

Gabor Fodor, Marcos V. Conde

Deforestation estimation and fire detection in the Amazon forest poses a significant challenge due to the vast size of the area and the limited accessibility. However, these are crucial problems that lead to severe environmental consequences, including climate change, global warming, and biodiversity loss. To effectively address this problem, multimodal satellite imagery and remote sensing offer a promising solution for estimating deforestation and detecting wildfire in the Amazonia region. This research paper introduces a new curated dataset and a deep learning-based approach to solve these problems using convolutional neural networks (CNNs) and comprehensive data processing techniques. Our dataset includes curated images and diverse channel bands from Sentinel, Landsat, VIIRS, and MODIS satellites. We design the dataset considering different spatial and temporal resolution requirements. Our method successfully achieves high-precision deforestation estimation and burned area detection on unseen images from the region. Our code, models and dataset are open source: https://github.com/h2oai/cvpr-multiearth-deforestation-segmentation

Federico Boccardi, Hossein Shokri-Ghadikolaei, Gabor Fodor et al.

Motivated by the intrinsic characteristics of mmWave technologies, we discuss the possibility of an authorization regime that allows spectrum sharing between multiple operators, also referred to as spectrum pooling. In particular, considering user rate as the performance measure, we assess the benefit of coordination among the networks of different operators, study the impact of beamforming both at the base stations and at the user terminals, and analyze the pooling performance at different frequency carriers. We also discuss the enabling spectrum mechanisms, architectures, and protocols required to make spectrum pooling work in real networks. Our initial results show that, from a technical perspective, spectrum pooling at mmWave has the potential for a more efficient spectrum use than a traditional exclusive spectrum allocation to a single operator. However, further studies are needed in order to reach a thorough understanding of this matter, and we hope that this paper will help stimulate further research in this area.

Fazal-E Asim, André L. F. de Almeida, Bruno Sokal et al.

As beyond-diagonal reconfigurable intelligent surfaces (BD-RISs) gain increasing attention in high-frequency wireless communications, accurate and scalable channel-estimation methods become essential. This paper develops a parametric channel-estimation and beamforming framework that deconstructs the composite BD-RIS channel into its generating directional factors, revealing the tensor structure induced jointly by propagation geometry and beyond-diagonal scattering. We propose two tensor-based estimators: Fourth-Order Tucker Channel Estimation (FORTE), which models the partially structured channel as a fourth-order Tucker tensor, and Fourth-Order PARAFAC Channel Estimation (FORPE), which captures the fully structured channel through a fourth-order PARAFAC model. By exploiting partial and full channel geometry, the proposed methods achieve higher estimation accuracy than Least Squares and Block Tucker Kronecker Factorization benchmarks. In particular, FORTE outperforms FORPE due to its more compact representation, attaining an NMSE of about 10^{-4} at 5 dB SNR. In contrast, FORPE provides essentially unique estimates of the composite-channel factor matrices, whereas FORTE identifies their subspaces. The proposed deconstruction also provides a structured representation useful for sensing-oriented parameter extraction and tensor-structured system optimization. Finally, the Tensor Optimization Framework for Beamforming, Combining, and Scattering (TenFormer) achieves spectral efficiency comparable to the benchmark design while significantly reducing computational complexity through parallel tensor-structured optimization.

Oscar Stenhammar, Gabor Fodor, Carlo Fischione

The performance of modern wireless communications systems depends critically on the quality of the available channel state information (CSI) at the transmitter and receiver. Several previous works have proposed concepts and algorithms that help maintain high quality CSI even in the presence of high mobility and channel aging, such as temporal prediction schemes that employ neural networks. However, it is still unclear which neural network-based scheme provides the best performance in terms of prediction quality, training complexity and practical feasibility. To investigate such a question, this paper first provides an overview of state-of-the-art neural networks applicable to channel prediction and compares their performance in terms of prediction quality. Next, a new comparative analysis is proposed for four promising neural networks with different prediction horizons. The well-known tapped delay channel model recommended by the Third Generation Partnership Program is used for a standardized comparison among the neural networks. Based on this comparative evaluation, the advantages and disadvantages of each neural network are discussed and guidelines for selecting the best-suited neural network in channel prediction applications are given.

Shashi Kant, José Mairton B. da Silva, Gabor Fodor et al.

Federated learning (FL) has emerged as an instance of distributed machine learning paradigm that avoids the transmission of data generated on the users' side. Although data are not transmitted, edge devices have to deal with limited communication bandwidths, data heterogeneity, and straggler effects due to the limited computational resources of users' devices. A prominent approach to overcome such difficulties is FedADMM, which is based on the classical two-operator consensus alternating direction method of multipliers (ADMM). The common assumption of FL algorithms, including FedADMM, is that they learn a global model using data only on the users' side and not on the edge server. However, in edge learning, the server is expected to be near the base station and have direct access to rich datasets. In this paper, we argue that leveraging the rich data on the edge server is much more beneficial than utilizing only user datasets. Specifically, we show that the mere application of FL with an additional virtual user node representing the data on the edge server is inefficient. We propose FedTOP-ADMM, which generalizes FedADMM and is based on a three-operator ADMM-type technique that exploits a smooth cost function on the edge server to learn a global model parallel to the edge devices. Our numerical experiments indicate that FedTOP-ADMM has substantial gain up to 33\% in communication efficiency to reach a desired test accuracy with respect to FedADMM, including a virtual user on the edge server.

Pascal Pfeiffer, Philipp Singer, Yauhen Babakhin et al.

We present H2O-Danube3, a series of small language models consisting of H2O-Danube3-4B, trained on 6T tokens and H2O-Danube3-500M, trained on 4T tokens. Our models are pre-trained on high quality Web data consisting of primarily English tokens in three stages with different data mixes before final supervised tuning for chat version. The models exhibit highly competitive metrics across a multitude of academic, chat, and fine-tuning benchmarks. Thanks to its compact architecture, H2O-Danube3 can be efficiently run on a modern smartphone, enabling local inference and rapid processing capabilities even on mobile devices. We make all models openly available under Apache 2.0 license further democratizing LLMs to a wider audience economically.

Amirreza Kazemi, Seyed Mohammad Azimi-Abarghouyi, Gabor Fodor et al.

Hierarchical federated learning (HFL) has emerged as a key architecture for large-scale wireless and Internet of Things systems, where devices communicate with nearby edge servers before reaching the cloud. In these environments, uplink bandwidth and latency impose strict communication limits, thereby making aggressive gradient compression essential. One-bit methods such as sign-based stochastic gradient descent (SignSGD) offer an attractive solution in flat federated settings, but existing theory and algorithms do not naturally extend to hierarchical settings. In particular, the interaction between majority-vote aggregation at the edge layer and model aggregation at the cloud layer, and its impact on end-to-end performance, remains unknown. To bridge this gap, we propose a highly communication-efficient sign-based HFL framework and develop its corresponding formulation for nonconvex learning, where devices send only signed stochastic gradients, edge servers combine them through majority-vote, and the cloud periodically averages the obtained edge models, while utilizing downlink quantization to broadcast the global model. We introduce the resulting scalable HFL algorithm, HierSignSGD, and provide the convergence analysis for SignSGD in a hierarchical setting. Our core technical contribution is a characterization of how biased sign compression, two-level aggregation intervals, and inter-cluster heterogeneity collectively affect convergence. Numerical experiments under homogeneous and heterogeneous data splits show that HierSignSGD, despite employing extreme compression, achieves accuracy comparable to or better than full-precision stochastic gradient descent while reducing communication cost in the process, and remains robust under aggressive downlink sparsification.

Philipp Singer, Pascal Pfeiffer, Yauhen Babakhin et al.

We present H2O-Danube, a series of small 1.8B language models consisting of H2O-Danube-1.8B, trained on 1T tokens, and the incremental improved H2O-Danube2-1.8B trained on an additional 2T tokens. Our models exhibit highly competitive metrics across a multitude of benchmarks and, as of the time of this writing, H2O-Danube2-1.8B achieves the top ranking on Open LLM Leaderboard for all models below the 2B parameter range. The models follow core principles of LLama 2 and Mistral, and we leverage and refine various techniques for pre-training large language models. We additionally release chat models trained with supervised fine-tuning followed by direct preference optimization. We make all models openly available under Apache 2.0 license further democratizing LLMs to a wider audience economically.

Samuele Peri, Alessio Russo, Gabor Fodor et al.

Link adaptation (LA) is an essential function in modern wireless communication systems that dynamically adjusts the transmission rate of a communication link to match time- and frequency-varying radio link conditions. However, factors such as user mobility, fast fading, imperfect channel quality information, and aging of measurements make the modeling of LA challenging. To bypass the need for explicit modeling, recent research has introduced online reinforcement learning (RL) approaches as an alternative to the more commonly used rule-based algorithms. Yet, RL-based approaches face deployment challenges, as training in live networks can potentially degrade real-time performance. To address this challenge, this paper considers offline RL as a candidate to learn LA policies with minimal effects on the network operation. We propose three LA designs based on batch-constrained deep Q-learning, conservative Q-learning, and decision transformer. Our results show that offline RL algorithms can match the performance of state-of-the-art online RL methods when data is collected with a proper behavioral policy.

Hossein S. Ghadikolaei, Hadi Ghauch, Gabor Fodor et al.

Inter-operator spectrum sharing in millimeter-wave bands has the potential of substantially increasing the spectrum utilization and providing a larger bandwidth to individual user equipment at the expense of increasing inter-operator interference. Unfortunately, traditional model-based spectrum sharing schemes make idealistic assumptions about inter-operator coordination mechanisms in terms of latency and protocol overhead, while being sensitive to missing channel state information. In this paper, we propose hybrid model-based and data-driven multi-operator spectrum sharing mechanisms, which incorporate model-based beamforming and user association complemented by data-driven model refinements. Our solution has the same computational complexity as a model-based approach but has the major advantage of having substantially less signaling overhead. We discuss how limited channel state information and quantized codebook-based beamforming affect the learning and the spectrum sharing performance. We show that the proposed hybrid sharing scheme significantly improves spectrum utilization under realistic assumptions on inter-operator coordination and channel state information acquisition.