Search

Fluorescence lifetime imaging (FLI) is a widely used technique in the biomedical field for measuring the decay times of fluorescent molecules, providing insights into metabolic states, protein interactions, and ligand-receptor bindings. However, its broader application in fast biological processes, such as dynamic activity monitoring, and clinical use, such as in guided surgery, is limited by long data acquisition times and computationally demanding data processing. While deep learning has reduced post-processing times, time-resolved data acquisition remains a bottleneck for real-time applications. To address this, we propose a method to achieve real-time FLI using an FPGA-based hardware accelerator. Specifically, we implemented a GRU-based sequence-to-sequence (Seq2Seq) model on an FPGA board compatible with time-resolved cameras. The GRU model balances accurate processing with the resource constraints of FPGAs, which have limited DSP units and BRAM. The limited memory and computational resources on the FPGA require efficient scheduling of operations and memory allocation to deploy deep learning models for low-latency applications. We address these challenges by using STOMP, a queue-based discrete-event simulator that automates and optimizes task scheduling and memory management on hardware. By integrating a GRU-based Seq2Seq model and its compressed version, called Seq2SeqLite, generated through knowledge distillation, we were able to process multiple pixels in parallel, reducing latency compared to sequential processing. We explore various levels of parallelism to achieve an optimal balance between performance and resource utilization. Our results indicate that the proposed techniques achieved a 17.7x and 52.0x speedup over manual scheduling for the Seq2Seq model and the Seq2SeqLite model, respectively., Comment: 7 pages, 6 figures

Morocco's Vision 2030, known as Maroc Digital 2030, aims to position the country as a regional leader in digital technology by boosting digital infrastructure, fostering innovation, and advancing digital skills. Complementing this initiative, the Pacte ESRI 2030 strategy, launched in 2023, seeks to transform the higher education, research, and innovation sectors by integrating state-of-the-art digital technologies. In alignment with these national strategies, this paper introduces BlockMEDC, a blockchain-based system for securing and managing Moroccan educational digital certificates. Leveraging Ethereum smart contracts and the InterPlanetary File System, BlockMEDC automates the issuance, management, and verification of academic credentials across Moroccan universities. The proposed system addresses key issues such as document authenticity, manual verification, and lack of interoperability, delivering a secure, transparent, and cost-effective solution that aligns with Morocco's digital transformation goals for the education sector.

In spite of the recent progress in speech processing, the majority of world languages and dialects remain uncovered. This situation only furthers an already wide technological divide, thereby hindering technological and socioeconomic inclusion. This challenge is largely due to the absence of datasets that can empower diverse speech systems. In this paper, we seek to mitigate this obstacle for a number of Arabic dialects by presenting Casablanca, a large-scale community-driven effort to collect and transcribe a multi-dialectal Arabic dataset. The dataset covers eight dialects: Algerian, Egyptian, Emirati, Jordanian, Mauritanian, Moroccan, Palestinian, and Yemeni, and includes annotations for transcription, gender, dialect, and code-switching. We also develop a number of strong baselines exploiting Casablanca. The project page for Casablanca is accessible at: www.dlnlp.ai/speech/casablanca.

Deep learning (DL) methods have been extensively applied to various image recovery problems, including magnetic resonance imaging (MRI) and computed tomography (CT) reconstruction. Beyond supervised models, other approaches have been recently explored including two key recent schemes: Deep Image Prior (DIP) that is an unsupervised scan-adaptive method that leverages the network architecture as implicit regularization but can suffer from noise overfitting, and diffusion models (DMs), where the sampling procedure of a pre-trained generative model is modified to allow sampling from the measurement-conditioned distribution through approximations. In this paper, we propose combining DIP and DMs for MRI and CT reconstruction, motivated by (i) the impact of the DIP network input and (ii) the use of DMs as diffusion purifiers (DPs). Specifically, we propose an unrolled procedure that iteratively optimizes the DIP network with a DM-refined adaptive input using a loss with data consistency and autoencoding terms. We term the approach unrolled Diffusion-Guided DIP (uDiG-DIP). Our experimental results demonstrate that uDiG-DIP achieves superior reconstruction results compared to leading DM-based baselines and the original DIP for MRI and CT tasks.

Diffusion models (DMs) are a class of generative models that allow sampling from a distribution learned over a training set. When applied to solving inverse imaging problems (IPs), the reverse sampling steps of DMs are typically modified to approximately sample from a measurement-conditioned distribution in the image space. However, these modifications may be unsuitable for certain settings (such as in the presence of measurement noise) and non-linear tasks, as they often struggle to correct errors from earlier sampling steps and generally require a large number of optimization and/or sampling steps. To address these challenges, we state three conditions for achieving measurement-consistent diffusion trajectories. Building on these conditions, we propose a new optimization-based sampling method that not only enforces the standard data manifold measurement consistency and forward diffusion consistency, as seen in previous studies, but also incorporates backward diffusion consistency that maintains a diffusion trajectory by optimizing over the input of the pre-trained model at every sampling step. By enforcing these conditions, either implicitly or explicitly, our sampler requires significantly fewer reverse steps. Therefore, we refer to our accelerated method as Step-wise Triple-Consistent Sampling (SITCOM). Compared to existing state-of-the-art baseline methods, under different levels of measurement noise, our extensive experiments across five linear and three non-linear image restoration tasks demonstrate that SITCOM achieves competitive or superior results in terms of standard image similarity metrics while requiring a significantly reduced run-time across all considered tasks.

Fluorescence lifetime imaging (FLI) is an important technique for studying cellular environments and molecular interactions, but its real-time application is limited by slow data acquisition, which requires capturing large time-resolved images and complex post-processing using iterative fitting algorithms. Deep learning (DL) models enable real-time inference, but can be computationally demanding due to complex architectures and large matrix operations. This makes DL models ill-suited for direct implementation on field-programmable gate array (FPGA)-based camera hardware. Model compression is thus crucial for practical deployment for real-time inference generation. In this work, we focus on compressing recurrent neural networks (RNNs), which are well-suited for FLI time-series data processing, to enable deployment on resource-constrained FPGA boards. We perform an empirical evaluation of various compression techniques, including weight reduction, knowledge distillation (KD), post-training quantization (PTQ), and quantization-aware training (QAT), to reduce model size and computational load while preserving inference accuracy. Our compressed RNN model, Seq2SeqLite, achieves a balance between computational efficiency and prediction accuracy, particularly at 8-bit precision. By applying KD, the model parameter size was reduced by 98\% while retaining performance, making it suitable for concurrent real-time FLI analysis on FPGA during data capture. This work represents a big step towards integrating hardware-accelerated real-time FLI analysis for fast biological processes., Comment: 8 pages, 2 figures

Brain-related research topics in artificial intelligence have recently gained popularity, particularly due to the expansion of what multimodal architectures can do from computer vision to natural language processing. Our main goal in this work is to explore the possibilities and limitations of these architectures in spoken text decoding from non-invasive fMRI recordings. Contrary to vision and textual data, fMRI data represent a complex modality due to the variety of brain scanners, which implies (i) the variety of the recorded signal formats, (ii) the low resolution and noise of the raw signals, and (iii) the scarcity of pretrained models that can be leveraged as foundation models for generative learning. These points make the problem of the non-invasive decoding of text from fMRI recordings very challenging. In this paper, we propose and end-to-end multimodal LLM for decoding spoken text from fMRI signals. The proposed architecture is founded on (i) an encoder derived from a specific transformer incorporating an augmented embedding layer for the encoder and a better-adjusted attention mechanism than that present in the state of the art, and (ii) a frozen large language model adapted to align the embedding of the input text and the encoded embedding of brain activity to decode the output text. A benchmark in performed on a corpus consisting of a set of interactions human-human and human-robot interactions where fMRI and conversational signals are recorded synchronously. The obtained results are very promising, as our proposal outperforms the evaluated models, and is able to generate text capturing more accurate semantics present in the ground truth. The implementation code is provided in https://github.com/Hmamouche/brain_decode., Comment: 15 pages, 4 figures

Sulfur vacancy defects mediate a wide range of optoelectronic properties in MoS2, with precise control of defect density allowing for tuneable optoelectronic devices. However, accurate measurement of defect density in monolayer and few-layer samples poses a challenge due to their small scattering cross-sections to photon or electron probes. Conventional lab-based techniques such as Raman and photoluminescence can infer approximate defect density in micro-scale samples via optoelectronic properties, but they require validation using stoichiometric beam-line XPS. We introduce an ultra-low energy (~64 meV) and non-intrusive lab-based technique to quantify the surface defect density in micron-scale monolayer MoS2. Here we show that a recently developed technique, helium atom micro-diffraction (referred to as scanning helium microscopy (SHeM) in literature), can be used to directly measure vacancy-type defect density in 2D materials by performing atom diffraction from a microscopic spot. SHeM uses a neutral, inert, and thermal energy probe of helium-4 atoms to measure ordered and disordered atom-surface scattering allowing the level of surface order to be inferred. The presented method enables rapid, non-damaging, and material-agnostic lab-based quantification of defect density in 2D materials, a crucial step towards the wider adoption of 2D semiconductors in devices., Comment: 15 pages, 7 figures

Integrating machine learning (ML) into customer service chatbots enhances their ability to understand and respond to user queries, ultimately improving service performance. However, they may appear artificial to some users and affecting customer experience. Hence, meticulous evaluation of ML models for each pipeline component is crucial for optimizing performance, though differences in functionalities can lead to unfair comparisons. In this paper, we present a tailored experimental evaluation approach for goal-oriented customer service chatbots with pipeline architecture, focusing on three key components: Natural Language Understanding (NLU), dialogue management (DM), and Natural Language Generation (NLG). Our methodology emphasizes individual assessment to determine optimal ML models. Specifically, we focus on optimizing hyperparameters and evaluating candidate models for NLU (utilizing BERT and LSTM), DM (employing DQN and DDQN), and NLG (leveraging GPT-2 and DialoGPT). The results show that for the NLU component, BERT excelled in intent detection whereas LSTM was superior for slot filling. For the DM component, the DDQN model outperformed DQN by achieving fewer turns, higher rewards, as well as greater success rates. For NLG, the large language model GPT-2 surpassed DialoGPT in BLEU, METEOR, and ROUGE metrics. These findings aim to provide a benchmark for future research in developing and optimizing customer service chatbots, offering valuable insights into model performance and optimal hyperparameters.

The recently observed very high energy (VHE) photons dubbed as GRB221009A by several terrestrial observatories such as LHAASO and Carpet-2 require a physics explanation beyond the standard model. Such energetic gamma ray bursts, originating from yet unknown very distance source at redshift z = 0.1505, would be directly scattererd by extragalactic background lights (EBL) rendering its improbable detection at the earth. We show that dark photon which resides in extra dimension would be able to resolve this issue when it oscillates resonantly with the photon in similar fashion with neutrino oscillation. We demonstrate that, for dark photon mass equals to 1 eV, the probability of GRB221009A photons with energy above 0.2 TeV to reach the earth lies in the range between $10^{-6} \, \text{to} \, 95\% $ for kinetic mixing values $10^{-14} \leq \epsilon \leq 10^{-12.5}$ allowed by current constraints., Comment: 10 pages, 3 figures

In many cases, adversarial attacks are based on specialized algorithms specifically dedicated to attacking automatic image classifiers. These algorithms perform well, thanks to an excellent ad hoc distribution of initial attacks. However, these attacks are easily detected due to their specific initial distribution. We therefore consider other black-box attacks, inspired from generic black-box optimization tools, and in particular the log-normal algorithm. We apply the log-normal method to the attack of fake detectors, and get successful attacks: importantly, these attacks are not detected by detectors specialized on classical adversarial attacks. Then, combining these attacks and deep detection, we create improved fake detectors., Comment: log-normal mutations and their use in detecting surreptitious fake images

The rapid advancement of intelligent technology has led to the development of optimization algorithms that leverage natural behaviors to address complex issues. Among these, the Rat Swarm Optimizer (RSO), inspired by rats' social and behavioral characteristics, has demonstrated potential in various domains, although its convergence precision and exploration capabilities are limited. To address these shortcomings, this study introduces the Modified Rat Swarm Optimizer (MRSO), designed to enhance the balance between exploration and exploitation. MRSO incorporates unique modifications to improve search efficiency and durability, making it suitable for challenging engineering problems such as welded beam, pressure vessel, and gear train design. Extensive testing with classical benchmark functions shows that MRSO significantly improves performance, avoiding local optima and achieving higher accuracy in six out of nine multimodal functions and in all seven fixed-dimension multimodal functions. In the CEC 2019 benchmarks, MRSO outperforms the standard RSO in six out of ten functions, demonstrating superior global search capabilities. When applied to engineering design problems, MRSO consistently delivers better average results than RSO, proving its effectiveness. Additionally, we compared our approach with eight recent and well-known algorithms using both classical and CEC-2019 bench-marks. MRSO outperforms each of these algorithms, achieving superior results in six out of 23 classical benchmark functions and in four out of ten CEC-2019 benchmark functions. These results further demonstrate MRSO's significant contributions as a reliable and efficient tool for optimization tasks in engineering applications., Comment: 36 pages

Traditional query answering over knowledge graphs -- or broadly over relational data -- is one of the most fundamental problems in data management. Motivated by the incompleteness of modern knowledge graphs, a new setup for query answering has emerged, where the goal is to predict answers that do not necessarily appear in the knowledge graph, but are present in its completion. In this work, we propose AnyCQ, a graph neural network model that can classify answers to any conjunctive query on any knowledge graph, following training. At the core of our framework lies a graph neural network model trained using a reinforcement learning objective to answer Boolean queries. Our approach and problem setup differ from existing query answering studies in multiple dimensions. First, we focus on the problem of query answer classification: given a query and a set of possible answers, classify these proposals as true or false relative to the complete knowledge graph. Second, we study the problem of query answer retrieval: given a query, retrieve an answer to the query relative to the complete knowledge graph or decide that no correct solutions exist. Trained on simple, small instances, AnyCQ can generalize to large queries of arbitrary structure, reliably classifying and retrieving answers to samples where existing approaches fail, which is empirically validated on new and challenging benchmarks. Furthermore, we demonstrate that our AnyCQ models effectively transfer to out-of-distribution knowledge graphs, when equipped with a relevant link predictor, highlighting their potential to serve as a general engine for query answering.

For run sizes that are a multiple of four, the literature offers many two-level designs that are D- and A-optimal for the main-effects model and minimize the aliasing between main effects and interaction effects and among interaction effects. For run sizes that are not a multiple of four, no conclusive results are known. In this paper, we propose two algorithms that generate all non-isomorphic D- and A-optimal main-effects designs for run sizes that are one and two more than a multiple of four. We enumerate all such designs for run sizes up to 18, report the numbers of designs we obtained, and identify those that minimize the aliasing between main effects and interaction effects and among interaction effects. Finally, we compare the minimally aliased designs we found with benchmark designs from the literature.

In this paper, a space-air-ground quantum (SPARQ) network is developed as a means for providing a seamless on-demand entanglement distribution. The node mobility in SPARQ poses significant challenges to entanglement routing. Existing quantum routing algorithms focus on stationary ground nodes and utilize link distance as an optimality metric, which is unrealistic for dynamic systems like SPARQ. Moreover, in contrast to the prior art that assumes homogeneous nodes, SPARQ encompasses heterogeneous nodes with different functionalities further complicates the entanglement distribution. To solve the entanglement routing problem, a deep reinforcement learning (RL) framework is proposed and trained using deep Q-network (DQN) on multiple graphs of SPARQ to account for the network dynamics. Subsequently, an entanglement distribution policy, third-party entanglement distribution (TPED), is proposed to establish entanglement between communication parties. A realistic quantum network simulator is designed for performance evaluation. Simulation results show that the TPED policy improves entanglement fidelity by 3% and reduces memory consumption by 50% compared with benchmark. The results also show that the proposed DQN algorithm improves the number of resolved teleportation requests by 39% compared with shortest path baseline and the entanglement fidelity by 2% compared with an RL algorithm that is based on long short-term memory (LSTM). It also improved entanglement fidelity by 6% and 9% compared with two state-of-the-art benchmarks. Moreover, the entanglement fidelity is improved by 15% compared with DQN trained on a snapshot of SPARQ. Additionally, SPARQ enhances the average entanglement fidelity by 23.5% compared with existing networks spanning only space and ground layers.

Voice conversion (VC) aims to modify the speaker's identity while preserving the linguistic content. Commonly, VC methods use an encoder-decoder architecture, where disentangling the speaker's identity from linguistic information is crucial. However, the disentanglement approaches used in these methods are limited as the speaker features depend on the phonetic content of the utterance, compromising disentanglement. This dependency is amplified with attention-based methods. To address this, we introduce a novel masking mechanism in the input before speaker encoding, masking certain discrete speech units that correspond highly with phoneme classes. Our work aims to reduce the phonetic dependency of speaker features by restricting access to some phonetic information. Furthermore, since our approach is at the input level, it is applicable to any encoder-decoder based VC framework. Our approach improves disentanglement and conversion performance across multiple VC methods, showing significant effectiveness, particularly in attention-based method, with 44% relative improvement in objective intelligibility., Comment: Accepted to IEEE SLT 2024

Advanced millimeter-wave software-defined array (SDA) platforms, or testbeds at affordable costs and high performance are essential for the wireless community. In this paper, we present a low-cost, portable, and programmable SDA that allows for accessible research and experimentation in real time. The proposed platform is based on a 16-element phased-array transceiver operating across 24-29.5 GHz, integrated with a radio-frequency system-on-chip board that provides data conversion and baseband signal-processing capabilities. All radio-communication parameters and phased-array beam configurations are controlled through a high-level application program interface. We present measurements evaluating the beamforming and communication link performance. Our experimental results validate that the SDA has a beam scan range of -45 to +45 degrees (azimuth), a 3 dB beamwidth of 20 degrees, and support up to a throughput of 1.613 Gb/s using 64-QAM. The signal-to-noise ratio is as high as 30 dB at short-range distances when the transmit and receive beams are aligned.

ANNZ is a fast and simple algorithm which utilises artificial neural networks (ANNs), it was known as one of the pioneers of machine learning approaches to photometric redshift estimation decades ago. We enhanced the algorithm by introducing new activation functions like tanh, softplus, SiLU, Mish and ReLU variants; its new performance is then vigorously tested on legacy samples like the Luminous Red Galaxy (LRG) and Stripe-82 samples from SDSS, as well as modern galaxy samples like the Physics of the Accelerating Universe Survey (PAUS). This work focuses on testing the robustness of activation functions with respect to the choice of ANN architectures, particularly on its depth and width, in the context of galaxy photometric redshift estimation. Our upgraded algorithm, which we named ANNZ+, shows that the tanh and Leaky ReLU activation functions provide more consistent and stable results across deeper and wider architectures with > 1 per cent improvement in root-mean-square error ($\sigma_{\textrm{RMS}}$) and 68th percentile error ($\sigma_{68}$) when tested on SDSS data sets. While assessing its capabilities in handling high dimensional inputs, we achieved an improvement of 11 per cent in $\sigma_{\textrm{RMS}}$ and 6 per cent in $\sigma_{68}$ with the tanh activation function when tested on the 40-narrowband PAUS dataset; it even outperformed ANNZ2, its supposed successor, by 44 per cent in $\sigma_{\textrm{RMS}}$. This justifies the effort to upgrade the 20-year-old ANNZ, allowing it to remain viable and competitive within the photo-z community today. The updated algorithm ANNZ+ is publicly available at https://github.com/imdadmpt/ANNzPlus., Comment: 37 pages, 9 figures, submitted to JCAP

Let $\mathscr{B}_n = \{ \pm x_1, \pm x_2, \pm x_3, \cdots, \pm x_{n-1}, x_n \}$ where $n>1$ is fixed, $x_i \in \mathbb{R}^+$, $i = 1, 2, 3, \cdots, n$ and $x_1 < x_2 < x_3 < \cdots < x_n$. Let $\phi(\mathscr{B}_n)$ be the set of all non-empty subsets $S = \{u_1, u_2,\cdots, u_t\}$ of $\mathscr{B}_n$ such that $|u_1|<|u_2|<\cdots <|u_{t-1}|

Detecting and classifying suspicious or malicious domain names and URLs is fundamental task in cybersecurity. To leverage such indicators of compromise, cybersecurity vendors and practitioners often maintain and update blacklists of known malicious domains and URLs. However, blacklists frequently fail to identify emerging and obfuscated threats. Over the past few decades, there has been significant interest in developing machine learning models that automatically detect malicious domains and URLs, addressing the limitations of blacklists maintenance and updates. In this paper, we introduce DomURLs_BERT, a pre-trained BERT-based encoder adapted for detecting and classifying suspicious/malicious domains and URLs. DomURLs_BERT is pre-trained using the Masked Language Modeling (MLM) objective on a large multilingual corpus of URLs, domain names, and Domain Generation Algorithms (DGA) dataset. In order to assess the performance of DomURLs_BERT, we have conducted experiments on several binary and multi-class classification tasks involving domain names and URLs, covering phishing, malware, DGA, and DNS tunneling. The evaluations results show that the proposed encoder outperforms state-of-the-art character-based deep learning models and cybersecurity-focused BERT models across multiple tasks and datasets. The pre-training dataset, the pre-trained DomURLs_BERT encoder, and the experiments source code are publicly available.

Click-through-rate (CTR) prediction plays an important role in online advertising and ad recommender systems. In the past decade, maximizing CTR has been the main focus of model development and solution creation. Therefore, researchers and practitioners have proposed various models and solutions to enhance the effectiveness of CTR prediction. Most of the existing literature focuses on capturing either implicit or explicit feature interactions. Although implicit interactions are successfully captured in some studies, explicit interactions present a challenge for achieving high CTR by extracting both low-order and high-order feature interactions. Unnecessary and irrelevant features may cause high computational time and low prediction performance. Furthermore, certain features may perform well with specific predictive models while underperforming with others. Also, feature distribution may fluctuate due to traffic variations. Most importantly, in live production environments, resources are limited, and the time for inference is just as crucial as training time. Because of all these reasons, feature selection is one of the most important factors in enhancing CTR prediction model performance. Simple filter-based feature selection algorithms do not perform well and they are not sufficient. An effective and efficient feature selection algorithm is needed to consistently filter the most useful features during live CTR prediction process. In this paper, we propose a heuristic algorithm named Neighborhood Search with Heuristic-based Feature Selection (NeSHFS) to enhance CTR prediction performance while reducing dimensionality and training time costs. We conduct comprehensive experiments on three public datasets to validate the efficiency and effectiveness of our proposed solution.

Using the approach derived by Caron-Huot and Gale from the BDMPS-Z formalism, we obtain medium-induced parton splitting rates in a Bjorken expanding QCD matter. We compare the rate with the case of a static medium and investigate the impact of the medium expansion on the splitting rate. We also consider the leading order in the opacity expansion and the harmonic oscillator solution. These approximations are recovered in the limits of small and large formation times, respectively., Comment: 8 pages, 4 figures

Integrating image and text data through multi-modal learning has emerged as a new approach in medical imaging research, following its successful deployment in computer vision. While considerable efforts have been dedicated to establishing medical foundation models and their zero-shot transfer to downstream tasks, the popular few-shot setting remains relatively unexplored. Following on from the currently strong emergence of this setting in computer vision, we introduce the first structured benchmark for adapting medical vision-language models (VLMs) in a strict few-shot regime and investigate various adaptation strategies commonly used in the context of natural images. Furthermore, we evaluate a simple generalization of the linear-probe adaptation baseline, which seeks an optimal blending of the visual prototypes and text embeddings via learnable class-wise multipliers. Surprisingly, such a text-informed linear probe yields competitive performances in comparison to convoluted prompt-learning and adapter-based strategies, while running considerably faster and accommodating the black-box setting. Our extensive experiments span three different medical modalities and specialized foundation models, nine downstream tasks, and several state-of-the-art few-shot adaptation methods. We made our benchmark and code publicly available to trigger further developments in this emergent subject: \url{https://github.com/FereshteShakeri/few-shot-MedVLMs}., Comment: MICCAI 2024 (Spotlight) - Code is available at https://github.com/FereshteShakeri/few-shot-MedVLMs.git

Deep learning models have been shown to be a powerful solution for Time Series Classification (TSC). State-of-the-art architectures, while producing promising results on the UCR and the UEA archives , present a high number of trainable parameters. This can lead to long training with high CO2 emission, power consumption and possible increase in the number of FLoating-point Operation Per Second (FLOPS). In this paper, we present a new architecture for TSC, the Light Inception with boosTing tEchnique (LITE) with only 2.34% of the number of parameters of the state-of-the-art InceptionTime model, while preserving performance. This architecture, with only 9, 814 trainable parameters due to the usage of DepthWise Separable Convolutions (DWSC), is boosted by three techniques: multiplexing, custom filters, and dilated convolution. The LITE architecture, trained on the UCR, is 2.78 times faster than InceptionTime and consumes 2.79 times less CO2 and power. To evaluate the performance of the proposed architecture on multivariate time series data, we adapt LITE to handle multivariate time series, we call this version LITEMV. To bring theory into application, we also conducted experiments using LITEMV on multivariate time series representing human rehabilitation movements, showing that LITEMV not only is the most efficient model but also the best performing for this application on the Kimore dataset, a skeleton based human rehabilitation exercises dataset. Moreover, to address the interpretability of LITEMV, we present a study using Class Activation Maps to understand the classification decision taken by the model during evaluation.

The development of vision-language models (VLMs) for histo-pathology has shown promising new usages and zero-shot performances. However, current approaches, which decompose large slides into smaller patches, focus solely on inductive classification, i.e., prediction for each patch is made independently of the other patches in the target test data. We extend the capability of these large models by introducing a transductive approach. By using text-based predictions and affinity relationships among patches, our approach leverages the strong zero-shot capabilities of these new VLMs without any additional labels. Our experiments cover four histopathology datasets and five different VLMs. Operating solely in the embedding space (i.e., in a black-box setting), our approach is highly efficient, processing $10^5$ patches in just a few seconds, and shows significant accuracy improvements over inductive zero-shot classification. Code available at https://github.com/FereshteShakeri/Histo-TransCLIP.

We propose a test of the significance of a variable appearing on the Lasso path and use it in a procedure for selecting one of the models of the Lasso path, controlling the Family-Wise Error Rate. Our null hypothesis depends on a set A of already selected variables and states that it contains all the active variables. We focus on the regularization parameter value from which a first variable outside A is selected. As the test statistic, we use this quantity's conditional p-value, which we define conditional on the non-penalized estimated coefficients of the model restricted to A. We estimate this by simulating outcome vectors and then calibrating them on the observed outcome's estimated coefficients. We adapt the calibration heuristically to the case of generalized linear models in which it turns into an iterative stochastic procedure. We prove that the test controls the risk of selecting a false positive in linear models, both under the null hypothesis and, under a correlation condition, when A does not contain all active variables. We assess the performance of our procedure through extensive simulation studies. We also illustrate it in the detection of exposures associated with drug-induced liver injuries in the French pharmacovigilance database.

Vision-Language Models for remote sensing have shown promising uses thanks to their extensive pretraining. However, their conventional usage in zero-shot scene classification methods still involves dividing large images into patches and making independent predictions, i.e., inductive inference, thereby limiting their effectiveness by ignoring valuable contextual information. Our approach tackles this issue by utilizing initial predictions based on text prompting and patch affinity relationships from the image encoder to enhance zero-shot capabilities through transductive inference, all without the need for supervision and at a minor computational cost. Experiments on 10 remote sensing datasets with state-of-the-art Vision-Language Models demonstrate significant accuracy improvements over inductive zero-shot classification. Our source code is publicly available on Github: https://github.com/elkhouryk/RS-TransCLIP

Synthesizing the voices of unseen speakers is a persisting challenge in multi-speaker text-to-speech (TTS). Most multi-speaker TTS models rely on modeling speaker characteristics through speaker conditioning during training. Modeling unseen speaker attributes through this approach has necessitated an increase in model complexity, which makes it challenging to reproduce results and improve upon them. We design a simple alternative to this. We propose SelectTTS, a novel method to select the appropriate frames from the target speaker and decode using frame-level self-supervised learning (SSL) features. We show that this approach can effectively capture speaker characteristics for unseen speakers, and achieves comparable results to other multi-speaker TTS frameworks in both objective and subjective metrics. With SelectTTS, we show that frame selection from the target speaker's speech is a direct way to achieve generalization in unseen speakers with low model complexity. We achieve better speaker similarity performance than SOTA baselines XTTS-v2 and VALL-E with over an 8x reduction in model parameters and a 270x reduction in training data., Comment: Submitted to IEEE Signal Processing Letters

Several quantum and classical Monte Carlo algorithms for Betti Number Estimation (BNE) on clique complexes have recently been proposed, though it is unclear how their performances compare. We review these algorithms, emphasising their common Monte Carlo structure within a new modular framework. This framework allows us to directly compare these algorithms by calculating upper bounds on the minimum number of samples needed for convergence. By recombining the different modules, we create a new quantum algorithm with an exponentially-improved dependence in the sample complexity. We run classical simulations to verify convergence within the theoretical bounds and observe the predicted exponential separation, even though empirical convergence occurs substantially earlier than the conservative theoretical bounds.

In recent years, the evolution of machine learning techniques has significantly impacted the field of intrusion detection, particularly within the context of the Internet of Things (IoT). As IoT networks expand, the need for robust security measures to counteract potential threats has become increasingly critical. This paper introduces a hybrid Intrusion Detection System (IDS) that synergistically combines Kolmogorov-Arnold Networks (KANs) with the XGBoost algorithm. Our proposed IDS leverages the unique capabilities of KANs, which utilize learnable activation functions to model complex relationships within data, alongside the powerful ensemble learning techniques of XGBoost, known for its high performance in classification tasks. This hybrid approach not only enhances the detection accuracy but also improves the interpretability of the model, making it suitable for dynamic and intricate IoT environments. Experimental evaluations demonstrate that our hybrid IDS achieves an impressive detection accuracy exceeding 99% in distinguishing between benign and malicious activities. Additionally, we were able to achieve F1 scores, precision, and recall that exceeded 98%. Furthermore, we conduct a comparative analysis against traditional Multi-Layer Perceptron (MLP) networks, assessing performance metrics such as Precision, Recall, and F1-score. The results underscore the efficacy of integrating KANs with XGBoost, highlighting the potential of this innovative approach to significantly strengthen the security framework of IoT networks., Comment: To be presented at the 11th International Symposium on Networks, Computers and Communications (ISNCC'24) will be held in Washington DC- USA, from October 22 to 25, 2024. Accepted (6 pages and 5 figures)

With the growing Deaf and Hard of Hearing population worldwide and the persistent shortage of certified sign language interpreters, there is a pressing need for an efficient, signs-driven, integrated end-to-end translation system, from sign to gloss to text and vice-versa. There has been a wealth of research on machine translations and related reviews. However, there are few works on sign language machine translation considering the particularity of the language being continuous and dynamic. This paper aims to address this void, providing a retrospective analysis of the temporal evolution of sign language machine translation algorithms and a taxonomy of the Transformers architectures, the most used approach in language translation. We also present the requirements of a real-time Quality-of-Service sign language ma-chine translation system underpinned by accurate deep learning algorithms. We propose future research directions for sign language translation systems.

A point cloud is a crucial geometric data structure utilized in numerous applications. The adoption of deep neural networks referred to as Point Cloud Neural Networks (PC- NNs), for processing 3D point clouds, has significantly advanced fields that rely on 3D geometric data to enhance the efficiency of tasks. Expanding the size of both neural network models and 3D point clouds introduces significant challenges in minimizing computational and memory requirements. This is essential for meeting the demanding requirements of real-world applications, which prioritize minimal energy consumption and low latency. Therefore, investigating redundancy in PCNNs is crucial yet challenging due to their sensitivity to parameters. Additionally, traditional pruning methods face difficulties as these networks rely heavily on weights and points. Nonetheless, our research reveals a promising phenomenon that could refine standard PCNN pruning techniques. Our findings suggest that preserving only the top p% of the highest magnitude weights is crucial for accuracy preservation. For example, pruning 99% of the weights from the PointNet model still results in accuracy close to the base level. Specifically, in the ModelNet40 dataset, where the base accuracy with the PointNet model was 87. 5%, preserving only 1% of the weights still achieves an accuracy of 86.8%. Codes are available in: https://github.com/apurba-nsu-rnd-lab/PCNN_Pruning, Comment: Accepted in BMVC 2024

Microservice architectures are increasingly used to modularize IoT applications and deploy them in distributed and heterogeneous edge computing environments. Over time, these microservice-based IoT applications are susceptible to performance anomalies caused by resource hogging (e.g., CPU or memory), resource contention, etc., which can negatively impact their Quality of Service and violate their Service Level Agreements. Existing research on performance anomaly detection for edge computing environments focuses on model training approaches that either achieve high accuracy at the expense of a time-consuming and resource-intensive training process or prioritize training efficiency at the cost of lower accuracy. To address this gap, while considering the resource constraints and the large number of devices in modern edge platforms, we propose two clustering-based model training approaches : (1) intra-cluster parameter transfer learning-based model training (ICPTL) and (2) cluster-level model training (CM). These approaches aim to find a trade-off between the training efficiency of anomaly detection models and their accuracy. We compared the models trained under ICPTL and CM to models trained for specific devices (most accurate, least efficient) and a single general model trained for all devices (least accurate, most efficient). Our findings show that the model accuracy of ICPTL is comparable to that of the model per device approach while requiring only 40% of the training time. In addition, CM further improves training efficiency by requiring 23% less training time and reducing the number of trained models by approximately 66% compared to ICPTL, yet achieving a higher accuracy than a single general model.

Recent studies show that large language models (LLMs) struggle with technical standards in telecommunications. We propose a fine-tuned retrieval-augmented generation (RAG) system based on the Phi-2 small language model (SLM) to serve as an oracle for communication networks. Our developed system leverages forward-looking semantic chunking to adaptively determine parsing breakpoints based on embedding similarity, enabling effective processing of diverse document formats. To handle the challenge of multiple similar contexts in technical standards, we employ a re-ranking algorithm to prioritize the most relevant retrieved chunks. Recognizing the limitations of Phi-2's small context window, we implement a recent technique, namely SelfExtend, to expand the context window during inference, which not only boosts the performance but also can accommodate a wider range of user queries and design requirements from customers to specialized technicians. For fine-tuning, we utilize the low-rank adaptation (LoRA) technique to enhance computational efficiency during training and enable effective fine-tuning on small datasets. Our comprehensive experiments demonstrate substantial improvements over existing question-answering approaches in the telecom domain, achieving performance that exceeds larger language models such as GPT-4 (which is about 880 times larger in size). This work presents a novel approach to leveraging SLMs for communication networks, offering a balance of efficiency and performance. This work can serve as a foundation towards agentic language models for networks., Comment: submitted to Proc. IEEE Globecom

In this article, we introduce a method for learning a capacity underlying a Sugeno integral according to training data based on systems of fuzzy relational equations. To the training data, we associate two systems of equations: a $\max-\min$ system and a $\min-\max$ system. By solving these two systems (in the case that they are consistent) using Sanchez's results, we show that we can directly obtain the extremal capacities representing the training data. By reducing the $\max-\min$ (resp. $\min-\max$) system of equations to subsets of criteria of cardinality less than or equal to $q$ (resp. of cardinality greater than or equal to $n-q$), where $n$ is the number of criteria, we give a sufficient condition for deducing, from its potential greatest solution (resp. its potential lowest solution), a $q$-maxitive (resp. $q$-minitive) capacity. Finally, if these two reduced systems of equations are inconsistent, we show how to obtain the greatest approximate $q$-maxitive capacity and the lowest approximate $q$-minitive capacity, using recent results to handle the inconsistency of systems of fuzzy relational equations.

The accuracy of mapping agricultural fields across large areas is steadily improving with high-resolution satellite imagery and deep learning (DL) models, even in regions where fields are small and geometrically irregular. However, developing effective DL models often requires large, expensive label datasets, typically available only for specific years or locations. This limits the ability to create annual maps essential for agricultural monitoring, as domain shifts occur between years and regions due to changes in farming practices and environmental conditions. The challenge is to design a model flexible enough to account for these shifts without needing yearly labels. While domain adaptation techniques or semi-supervised training are common solutions, we explored enhancing the model's generalization power. Our results indicate that a holistic approach is essential, combining methods to improve generalization. Specifically, using an area-based loss function, such as Tversky-focal loss (TFL), significantly improved predictions across multiple years. The use of different augmentation techniques helped to encode different types of invariance, particularly photometric augmentations encoded invariance to brightness changes, though they increased false positives. The combination of photometric augmentation, TFL loss, and MC-dropout produced the best results, although dropout alone led to more false negatives in subsequent year predictions. Additionally, the choice of input normalization had a significant impact, with the best results obtained when statistics were calculated either locally or across the entire dataset over all bands (lab and gab). We developed a workflow that enabled a U-Net model to generate effective multi-year crop maps over large areas. Our code, available at: https://github.com/agroimpacts/cnn-generalization-enhancement, will be regularly updated with improvements.

This thesis presents and evaluates the Dilated Convolution with Learnable Spacings (DCLS) method. Through various supervised learning experiments in the fields of computer vision, audio, and speech processing, the DCLS method proves to outperform both standard and advanced convolution techniques. The research is organized into several steps, starting with an analysis of the literature and existing convolution techniques that preceded the development of the DCLS method. We were particularly interested in the methods that are closely related to our own and that remain essential to capture the nuances and uniqueness of our approach. The cornerstone of our study is the introduction and application of the DCLS method to convolutional neural networks (CNNs), as well as to hybrid architectures that rely on both convolutional and visual attention approaches. DCLS is shown to be particularly effective in tasks such as classification, semantic segmentation, and object detection. Initially using bilinear interpolation, the study also explores other interpolation methods, finding that Gaussian interpolation slightly improves performance. The DCLS method is further applied to spiking neural networks (SNNs) to enable synaptic delay learning within a neural network that could eventually be transferred to so-called neuromorphic chips. The results show that the DCLS method stands out as a new state-of-the-art technique in SNN audio classification for certain benchmark tasks in this field. These tasks involve datasets with a high temporal component. In addition, we show that DCLS can significantly improve the accuracy of artificial neural networks for the multi-label audio classification task. We conclude with a discussion of the chosen experimental setup, its limitations, the limitations of our method, and our results., Comment: PhD Thesis

Our study presents a multifaceted approach to enhancing user interaction and content relevance in social media platforms through a federated learning framework. We introduce personalized GPT and Context-based Social Media LLM models, utilizing federated learning for privacy and security. Four client entities receive a base GPT-2 model and locally collected social media data, with federated aggregation ensuring up-to-date model maintenance. Subsequent modules focus on categorizing user posts, computing user persona scores, and identifying relevant posts from friends' lists. A quantifying social engagement approach, coupled with matrix factorization techniques, facilitates personalized content suggestions in real-time. An adaptive feedback loop and readability score algorithm also enhance the quality and relevance of content presented to users. Our system offers a comprehensive solution to content filtering and recommendation, fostering a tailored and engaging social media experience while safeguarding user privacy., Comment: This research paper is submitted to ASONAM 2024 conference on Advances in Social Networks Analysis and Mining and going to be published in Springer

Our paper introduces a novel approach to social network information retrieval and user engagement through a personalized chatbot system empowered by Federated Learning GPT. The system is designed to seamlessly aggregate and curate diverse social media data sources, including user posts, multimedia content, and trending news. Leveraging Federated Learning techniques, the GPT model is trained on decentralized data sources to ensure privacy and security while providing personalized insights and recommendations. Users interact with the chatbot through an intuitive interface, accessing tailored information and real-time updates on social media trends and user-generated content. The system's innovative architecture enables efficient processing of input files, parsing and enriching text data with metadata, and generating relevant questions and answers using advanced language models. By facilitating interactive access to a wealth of social network information, this personalized chatbot system represents a significant advancement in social media communication and knowledge dissemination., Comment: This research paper is submitted to ASONAM 2024 conference on Advances in Social Networks Analysis and Mining and going to be published in Springer

Dilated Convolution with Learnable Spacing (DCLS) is a recent advanced convolution method that allows enlarging the receptive fields (RF) without increasing the number of parameters, like the dilated convolution, yet without imposing a regular grid. DCLS has been shown to outperform the standard and dilated convolutions on several computer vision benchmarks. Here, we show that, in addition, DCLS increases the models' interpretability, defined as the alignment with human visual strategies. To quantify it, we use the Spearman correlation between the models' GradCAM heatmaps and the ClickMe dataset heatmaps, which reflect human visual attention. We took eight reference models - ResNet50, ConvNeXt (T, S and B), CAFormer, ConvFormer, and FastViT (sa 24 and 36) - and drop-in replaced the standard convolution layers with DCLS ones. This improved the interpretability score in seven of them. Moreover, we observed that Grad-CAM generated random heatmaps for two models in our study: CAFormer and ConvFormer models, leading to low interpretability scores. We addressed this issue by introducing Threshold-Grad-CAM, a modification built on top of Grad-CAM that enhanced interpretability across nearly all models. The code and checkpoints to reproduce this study are available at: https://github.com/rabihchamas/DCLS-GradCAM-Eval., Comment: Accepted at The Trustworthy AI Workshop, IJCAI 2024