Afonso Ferreira

Generative models have emerged as transformative tools in several domains, including computer vision and natural language processing, by learning data distributions and generating samples from them..... #compressing #Diffusion #latent #LDM #LDPruner #method #models #Nota #performancepreserving #present #pruning #Researchers #Structured

Deep dive into Artificial intelligence(AI) and what it means in our fields of expertise. Advancements in AI provide incredible potential to revolutionize the healthcare systems. A good example would be integration of AI in supply chain system this could improve lead times at demand and supply planning, a more accurate forecast, automated ordering systems based on the inventory management, supply chain transparency and optimization, reduced lead times at SDPs and could make supply chain more cost effective and sustainable. For now check out the link below on how I used AI to create a song in less than 20min. listen to it and appreciate how incredible AI is and what it can do in different respects of our lives enjoy it I know I did 😀. https://lnkd.in/dhTWmQ9v #Udio, #AI_Made_Music #AI_Tunes_Africa #InnovateWithAI and #DoHardThings. 🌍🎧 #ALX_AICE #ALX #AI_for_Africa

Today the most important article in chatgpt and "Can large language models pass official high-grade exams of the European Society of Neuroradiology courses? A direct comparison between OpenAI chatGPT 3.5, OpenAI GPT4 and Google Bard" - The study compared the performance of different Large Language Models (LLMs) including chatGPT, GPT4, and Google Bard in executing examinations from four different courses of the European Society of Neuroradiology (ESNR). - Exams related to anatomy/embryology, neuro-oncology, head and neck, and pediatrics were used as input data for the comparison. - Statistical analyses revealed a significant difference in accuracy between the LLMs, with GPT4 achieving the highest accuracy at 70%, followed by chatGPT 3.5 at 54% and Google Bard at 36%. - Pair-wise comparisons showed significant differences between the LLMs, with GPT4

📃Scientific paper: Efficient iterative methods for hyperparameter estimation in large-scale linear inverse problems Abstract: We study Bayesian methods for large-scale linear inverse problems, focusing on the challenging task of hyperparameter estimation. Typical hierarchical Bayesian formulations that follow a Markov Chain Monte Carlo approach are possible for small problems with very few hyperparameters but are not computationally feasible for problems with a very large number of unknown parameters. In this work, we describe an empirical Bayesian (EB) method to estimate hyperparameters that maximize the marginal posterior, i.e., the probability density of the hyperparameters conditioned on the data, and then we use the estimated values to compute the posterior of the inverse parameters. For problems where the computation of the square root and inverse of prior covariance matrices are not feasible, we describe an approach based on the generalized Golub-Kahan bidiagonalization to approximate the marginal posterior and seek hyperparameters that minimize the approximate marginal posterior. Numerical results from seismic and atmospheric tomography demonstrate the accuracy, robustness, and potential benefits of the proposed approach. Continued on ES/IODE ➡️ https://etcse.fr/dyB3c ------- If you find this interesting, feel free to follow, comment and share. We need your help to enhance our visibility, so that our platform continues to serve you.

👉🏼 Diagnosis in Bytes: Comparing the Diagnostic Accuracy of Google and ChatGPT 3.5 as an Educational Support Tool 🤓 Guilherme R Guimaraes 👇🏻 https://lnkd.in/eYfRGsCG 🔍 Focus on data insights: - Google's accuracy for commonly encountered urological conditions was 53.3%, with an additional 23.3% falling within a plausible range of differential diagnoses. - ChatGPT 3.5 outperformed Google with an accuracy of 86.6% in diagnosing prevalent conditions. - ChatGPT 3.5 identified the proper diagnosis in 16.6% of rare cases and offered a reasonable differential diagnosis in half of the cases of unusual disorders. 💡 Main outcomes and implications: - ChatGPT 3.5 demonstrated higher diagnostic accuracy than Google in both contexts. - ChatGPT 3.5 showed satisfactory accuracy when diagnosing common cases, but its performance in identifying rare conditions remains limited. 📚 Field significance: - Evaluating and comparing the diagnostic accuracy of digital tools like Google and ChatGPT 3.5 provides valuable insights into their potential as educational support tools in healthcare. - Understanding the strengths and limitations of these tools can aid in improving diagnostic processes and patient care. 🗄️: [#urology #diagnosticaccuracy #digitaltools #healthcare]

Table of basic properties This table shows some mathematical operations in the time domain and the corresponding effect in the Fourier series coefficients.  Linearity Time reversal / Frequency reversal Time conjugation Time reversal & conjugation Real part in time Imaginary part in time Real part in frequency Imaginary part in frequency Shift in time / Modulation in frequency Shift in frequency / Modulation in time Symmetry properties From this, various relationships are apparent, for example: The transform of a real-valued function (sRE + sRO) is the even symmetric function SRE + i SIO. Conversely, an even-symmetric transform implies a real-valued time-domain. The transform of an imaginary-valued function (i sIE + i sIO) is the odd symmetric function SRO + i SIE, and the converse is true. The transform of an even-symmetric function (sRE + i sIO) is the real-valued function SRE + SRO, and the converse is true. The transform of an odd-symmetric function (sRO + i sIE) is the imaginary-valued function i SIE + i SIO, and the converse is true. Riemannian manifolds Main articles: Laplace operator and Riemannian manifold If the domain is not a group, then there is no intrinsically defined convolution. However, if 𝑋 is a compact Riemannian manifold, it has a Laplace–Beltrami operator. The Laplace–Beltrami operator is the differential operator that corresponds to Laplace operator for the Riemannian manifold 𝑋 . Then, by analogy, one can consider heat equations on 𝑋 . Since Fourier arrived at his basis by attempting to solve the heat equation, the natural generalization is to use the eigensolutions of the Laplace–Beltrami operator as a basis. This generalizes Fourier series to spaces of the type 𝐿^2(𝑋), where 𝑋 is a Riemannian manifold. The Fourier series converges in ways similar to the [−𝜋,𝜋] case. A typical example is to take 𝑋 to be the sphere with the usual metric, in which case the Fourier basis consists of spherical harmonics.

👉🏼 Can large language models pass official high-grade exams of the European Society of Neuroradiology courses? A direct comparison between OpenAI chatGPT 3.5, OpenAI GPT4 and Google Bard 🤓 Gennaro D'Anna 👇🏻 https://lnkd.in/exSsRbPp 🔍 Focus on data insights: - Large language models (LLMs) performance comparison in executing examinations from four different courses of the European Society of Neuroradiology. - Evaluation of LLMs accuracy in subspecialty domains such as anatomy/embryology, neuro-oncology, head and neck, and pediatrics. 💡 Main outcomes and implications: - GPT4 demonstrated the highest accuracy at 70%, outperforming chatGPT 3.5 (54%) and Google Bard (36%). - Significant differences observed between the three LLMs in running official exams organized by ESNR. - Performance variations among LLMs were most pronounced in the head and neck subspecialty domain. 📚 Field significance: - Language model performance in specialized medical domains. - Implications for utilizing LLMs in educational assessment and examination settings. 🗄️: #Neuroradiology #LanguageModels #MedicalEducation #DataInsights

📃Scientific paper: TA-RNN: an Attention-based Time-aware Recurrent Neural Network Architecture for Electronic Health Records Abstract: Motivation: Electronic Health Records (EHR) represent a comprehensive resource of a patient's medical history. EHR are essential for utilizing advanced technologies such as deep learning (DL), enabling healthcare providers to analyze extensive data, extract valuable insights, and make precise and data-driven clinical decisions. DL methods such as Recurrent Neural Networks (RNN) have been utilized to analyze EHR to model disease progression and predict diagnosis. However, these methods do not address some inherent irregularities in EHR data such as irregular time intervals between clinical visits. Furthermore, most DL models are not interpretable. In this study, we propose two interpretable DL architectures based on RNN, namely Time-Aware RNN (TA-RNN) and TA-RNN-Autoencoder (TA-RNN-AE) to predict patient's clinical outcome in EHR at next visit and multiple visits ahead, respectively. To mitigate the impact of irregular time intervals, we propose incorporating time embedding of the elapsed times between visits. For interpretability, we propose employing a dual-level attention mechanism that operates between visits and features within each visit. Results: The results of the experiments conducted on Alzheimer's Disease Neuroimaging Initiative (ADNI) and National Alzheimer's Coordinating Center (NACC) datasets indicated superior performance of proposed models for predicting Alzheimer's Disease (AD) compared to state-of-the-art and baseline approaches based on F2 and sens... Continued on ES/IODE ➡️ https://etcse.fr/quzIc ------- If you find this interesting, feel free to follow, comment and share. We need your help to enhance our visibility, so that our platform continues to serve you. #alzheimer #science #health

📃Scientific paper: TA-RNN: an Attention-based Time-aware Recurrent Neural Network Architecture for Electronic Health Records Abstract: Motivation: Electronic Health Records (EHR) represent a comprehensive resource of a patient's medical history. EHR are essential for utilizing advanced technologies such as deep learning (DL), enabling healthcare providers to analyze extensive data, extract valuable insights, and make precise and data-driven clinical decisions. DL methods such as Recurrent Neural Networks (RNN) have been utilized to analyze EHR to model disease progression and predict diagnosis. However, these methods do not address some inherent irregularities in EHR data such as irregular time intervals between clinical visits. Furthermore, most DL models are not interpretable. In this study, we propose two interpretable DL architectures based on RNN, namely Time-Aware RNN (TA-RNN) and TA-RNN-Autoencoder (TA-RNN-AE) to predict patient's clinical outcome in EHR at next visit and multiple visits ahead, respectively. To mitigate the impact of irregular time intervals, we propose incorporating time embedding of the elapsed times between visits. For interpretability, we propose employing a dual-level attention mechanism that operates between visits and features within each visit. Results: The results of the experiments conducted on Alzheimer's Disease Neuroimaging Initiative (ADNI) and National Alzheimer's Coordinating Center (NACC) datasets indicated superior performance of proposed models for predicting Alzheimer's Disease (AD) compared to state-of-the-art and baseline approaches based on F2 and sens... Continued on ES/IODE ➡️ https://etcse.fr/quzIc ------- If you find this interesting, feel free to follow, comment and share. We need your help to enhance our visibility, so that our platform continues to serve you. #alzheimer #science #health

📃Scientific paper: TA-RNN: an Attention-based Time-aware Recurrent Neural Network Architecture for Electronic Health Records Abstract: Motivation: Electronic Health Records (EHR) represent a comprehensive resource of a patient's medical history. EHR are essential for utilizing advanced technologies such as deep learning (DL), enabling healthcare providers to analyze extensive data, extract valuable insights, and make precise and data-driven clinical decisions. DL methods such as Recurrent Neural Networks (RNN) have been utilized to analyze EHR to model disease progression and predict diagnosis. However, these methods do not address some inherent irregularities in EHR data such as irregular time intervals between clinical visits. Furthermore, most DL models are not interpretable. In this study, we propose two interpretable DL architectures based on RNN, namely Time-Aware RNN (TA-RNN) and TA-RNN-Autoencoder (TA-RNN-AE) to predict patient's clinical outcome in EHR at next visit and multiple visits ahead, respectively. To mitigate the impact of irregular time intervals, we propose incorporating time embedding of the elapsed times between visits. For interpretability, we propose employing a dual-level attention mechanism that operates between visits and features within each visit. Results: The results of the experiments conducted on Alzheimer's Disease Neuroimaging Initiative (ADNI) and National Alzheimer's Coordinating Center (NACC) datasets indicated superior performance of proposed models for predicting Alzheimer's Disease (AD) compared to state-of-the-art and baseline approaches based on F2 and sens... Continued on ES/IODE ➡️ https://etcse.fr/quzIc ------- If you find this interesting, feel free to follow, comment and share. We need your help to enhance our visibility, so that our platform continues to serve you. #alzheimer #science #health

📃Scientific paper: TA-RNN: an Attention-based Time-aware Recurrent Neural Network Architecture for Electronic Health Records Abstract: Motivation: Electronic Health Records (EHR) represent a comprehensive resource of a patient's medical history. EHR are essential for utilizing advanced technologies such as deep learning (DL), enabling healthcare providers to analyze extensive data, extract valuable insights, and make precise and data-driven clinical decisions. DL methods such as Recurrent Neural Networks (RNN) have been utilized to analyze EHR to model disease progression and predict diagnosis. However, these methods do not address some inherent irregularities in EHR data such as irregular time intervals between clinical visits. Furthermore, most DL models are not interpretable. In this study, we propose two interpretable DL architectures based on RNN, namely Time-Aware RNN (TA-RNN) and TA-RNN-Autoencoder (TA-RNN-AE) to predict patient's clinical outcome in EHR at next visit and multiple visits ahead, respectively. To mitigate the impact of irregular time intervals, we propose incorporating time embedding of the elapsed times between visits. For interpretability, we propose employing a dual-level attention mechanism that operates between visits and features within each visit. Results: The results of the experiments conducted on Alzheimer's Disease Neuroimaging Initiative (ADNI) and National Alzheimer's Coordinating Center (NACC) datasets indicated superior performance of proposed models for predicting Alzheimer's Disease (AD) compared to state-of-the-art and baseline approaches based on F2 and sens... Continued on ES/IODE ➡️ https://etcse.fr/quzIc ------- If you find this interesting, feel free to follow, comment and share. We need your help to enhance our visibility, so that our platform continues to serve you. #alzheimer #science #health