arXiv subjects

Marcelo Nogueira

Publications and source records attributed to Marcelo Nogueira.

8 recordsLinked to original sources

arXiv2026

Memory-Aware Multi-Agent Architecture with LLM-based Orchestration for Conversational Recommendation

Conversational recommender systems often treat user sessions in isolation, limiting their ability to capture how user intent evolves over time. While large language models (LLMs) have improved semantic understanding, they are typically used as isolated components, without addressing how memory, retrieval, and decision-making should be integrated at the system level. In this work, we propose a memory-aware multi-agent architecture for conversational recommendation, in which LLMs act as orchestration layers that coordinate intent refinement, query construction, and interaction flow. A key design principle is the integration of long-term memory as a first-class component, influencing both routing decisions and semantic query formation. We evaluate the system through cross-session memory experiments, multi-turn conversational simulations, and supervisor-driven interaction studies on the Amazon Reviews dataset. Results show that memory-enabled interactions recover up to 63\% of the gap toward explicit-query scenarios and prevent semantic degradation in extended dialogues. In addition, supervisory orchestration increases user acceptance by +28.9 percentage points, highlighting the importance of structured orchestration in guiding interaction and output quality. We also observe an exposure bottleneck between retrieved candidates and displayed recommendations. While LLM-reranking does not improve Hit@K, it consistently enhances semantic consistency and diversity. To capture these effects, we introduce the Semantic Consistency Score (SCS), a complementary evaluation measure. Overall, the results emphasize the importance of system-level design choices, suggesting that effective conversational recommendation benefits from explicit memory integration, orchestration, and evaluation beyond traditional ranking metrics.

cs.AI↗

arXiv2026

Towards Automatic Heart Murmur Characterization from Multi-spot Phonocardiograms

Cardiac auscultation traditionally detects the presence of a heart murmur, but clinicians further characterize it by timing, shape, pitch, quality, and grade. Most automatic phonocardiogram analysis systems stop at binary detection and do not estimate this richer description. In this work, we formulate automatic murmur characterization as five patient-level classification tasks and compare three multi-spot strategies on recordings from the aortic, pulmonary, mitral, and tricuspid auscultation locations. First, classical and deep-learning classifiers are trained independently at each location, and predictions are combined via a maximum-severity late-fusion rule. Second, a synchronous multi-input CNN jointly processes one representative segment per location for a single task. Third, a unified multi-input, multi-output CNN predicts all five attributes from one shared multi-spot representation. Using 735 patients from the CirCor DigiScope dataset with complete four-spot recordings, the best late-fusion configurations reach an unweighted mean sensitivity between 85.56% and 93.81% across the five tasks, with class-specific recalls between 80.00% and 100.00% for every relevant category. The multi-input CNN trails late fusion by 7-16 points depending on the task, while the multi-output model reaches values between 72.59% and 79.40%, estimating all five attributes from one shared representation. We further show that naively balancing training data by adding all minority-class segments degrades patient-level performance under the maximum-severity rule, since it introduces non-representative, noisy segments not filtered by the clustering-based segment selection used for evaluation. These results indicate multi-spot murmur characterization is feasible with both fusion-based and jointly learned architectures, outlining a path toward automatic, structured murmur reports complementing binary screening.

cs.LG↗

arXiv2026

Selection of Heart Sound Segments for Synchronous Classification of Multi-channel Heart Sounds

Cardiac auscultation remains the most cost-effective screening procedure for cardiovascular diseases, and requires listening at the four main auscultation spots. Despite this, automatic heart sound analysis algorithms mostly classify patients using a single heart sound (single-channel), or, when using more than one (multi-channel), analyze each channel individually. To our knowledge, no prior work classifies patients through the synchronous analysis of multi-channel heart sounds, following the procedure used by physicians. This motivates us to study whether synchronous multi-channel analysis outperforms single-channel approaches, and whether it holds an advantage over asynchronous multi-channel methods that analyze channels one by one, potentially by capturing inter-channel interference phenomena. To answer these questions, we introduce a selection algorithm that identifies optimal heart sound segments from each of the four auscultation spots, which are then fed into a multi-input CNN that classifies patients by analyzing the four selected sounds simultaneously. Our synchronous approach, combining the proposed selection algorithm with a multi-input CNN, achieves a superior overall accuracy of 96.5\%, a 9.1\% gain over the best-performing single-channel and asynchronous multi-channel methods. The benefit of the proposed segment selection strategy over random selection is confirmed by a paired statistical significance test ($p = 0.003$). These results were obtained on 735 patients from the CirCor DigiScope dataset with complete recordings from all four spots, and their scope and generalizability are discussed in light of this and other methodological considerations.

cs.LG↗

arXiv2026

Instantaneous analytic smoothing of rough data for the modified and cubic gKdV equations

We consider the $k$-generalized Korteweg-de Vries equation \begin{equation*} \partial_{t}v+\partial_{x}^{3}v +\partial_{x}(v^{k+1})=0, \qquad (t,x)\in\mathbb R\times\mathbb R, \qquad k\in\mathbb Z_{+}, \end{equation*} emphasizing the modified case $k=2$ and the cubic case $k=3$. We prove that solutions from low-regularity, possibly singular, data $u_{0}$ become real analytic in $(t,x)$ for all $t\neq0$, whenever $u_0$ satisfies a Nelson-type condition \begin{equation*} \sum_{k=0}^{\infty}\frac{α^{k}}{k!}\,\big\|(x\partial_x)^{k}u_0\big\|_{X}<\infty, \end{equation*} for some $α>0$. For the cubic equation, this includes data such as $u_0=x_{+}^λ$, singular at the origin; for mKdV even discontinuous data yield analytic solutions \emph{e.g} $u_{0}(x)=\sgn(x)e^{-x^{2}}$.For mKdV we work in the sharp well-posedness space $X=\widehat H^{r}_{s}(\mathbb R)$, $r\in(1,2]$, $s\geq\frac12-\frac1{2r}$, with $r=2$ recovering analyticity on $H^s(\mathbb R)$, $s\geq\frac14$, the best mKdV space in the sense of Kato; for the cubic equation we work in $X=H^{s}(\mathbb R)$, $s>-\frac16$, approaching the critical exponent $s=-\frac16$ from above. Analyticity thus holds on the largest known data class for which mKdV is well-posed, and on data approaching the corresponding threshold for the cubic equation, extending a known smoothing effect for KdV ($k=1$) to the modified and cubic nonlinearities and to a broader class of singular profiles, avoiding pseudo-differential calculus via Lorentz-space refinements replacing Bourgain-space localization. The mechanism is dispersive: analyticity is generated by the flow, symmetrically in time, and singular profiles become instantaneously analytic for $t\neq0$.

math.AP↗

arXiv2023

Local and global well-posedness for a quadratic Schrödinger system on spheres and Zoll manifolds

We consider the initial value problem (IVP) associated to a quadratic Schrödinger system \begin{equation*} \begin{cases} i \partial_{t} v \pm Δ_{g} v - v = ε_{1} u \bar{v}, & t \in \mathbb{R},\; x \in M, \\[2ex] i σ\partial_{t} u \pm Δ_{g} u - αu = \frac{ε_{2}}{2} v^{2}, & σ> 0, \;α\in \mathbb{R},\; ε_{i} \in \mathbb{C}\, (i = 1, 2),\\[2ex] (v(0), u(0)) = (v_0, u_0), \end{cases} \end{equation*} posed on a $d$-dimensional sphere $ \mathbb{S}^{d}$ or a compact Zoll manifold $M$. Considering $σ=\fracθβ$ with $θ, β\in \{n^2:n\in\mathbb{Z}\}$ we derive a bilinear Strichartz type estimate and use it to prove the local well-posedness results for given data $(v_0, u_0)\in H^s(M)\times H^s(M)$ whenever $s>\frac{1}{4}$ in the case $M = \mathbb{S}^{2}$ or a Zoll manifold, and $s > \frac{d - 2}{2}$ in the case $M = \mathbb{S}^{d}$ ($d \geq 3$) induced with the canonical metric. Moreover, in dimensions $2$ and $3$, we use a Gagliardo-Nirenberg type inequality to prove that the local solution can be extended globally in time whenever $s \geq 1$.

math.AP↗

arXiv2021

The CirCor DigiScope Dataset: From Murmur Detection to Murmur Classification

Cardiac auscultation is one of the most cost-effective techniques used to detect and identify many heart conditions. Computer-assisted decision systems based on auscultation can support physicians in their decisions. Unfortunately, the application of such systems in clinical trials is still minimal since most of them only aim to detect the presence of extra or abnormal waves in the phonocardiogram signal, i.e., only a binary ground truth variable (normal vs abnormal) is provided. This is mainly due to the lack of large publicly available datasets, where a more detailed description of such abnormal waves (e.g., cardiac murmurs) exists. To pave the way to more effective research on healthcare recommendation systems based on auscultation, our team has prepared the currently largest pediatric heart sound dataset. A total of 5282 recordings have been collected from the four main auscultation locations of 1568 patients, in the process, 215780 heart sounds have been manually annotated. Furthermore, and for the first time, each cardiac murmur has been manually annotated by an expert annotator according to its timing, shape, pitch, grading, and quality. In addition, the auscultation locations where the murmur is present were identified as well as the auscultation location where the murmur is detected more intensively. Such detailed description for a relatively large number of heart sounds may pave the way for new machine learning algorithms with a real-world application for the detection and analysis of murmur waves for diagnostic purposes.

q-bio.QM↗

arXiv2019

Local well-posedness for the quadratic Schrodinger equation in two-dimensional compact manifolds with boundary

We consider the quadractic NLS posed on a bidimensional compact Riemannian manifold $(M, g)$ with $ \partial M \neq \emptyset$. Using bilinear and gradient bilinear Strichartz estimates for Schrödinger operators in two-dimensional compact manifolds proved by J. Jiang in \cite{JIANG} we deduce a new evolution bilinear estimates. Consequently, using Bourgain's spaces, we obtain a local well-posedness result for given data $u_0\in H^s(M)$ whenever $s> \frac{2}{3}$ in such manifolds.

math.AP↗

arXiv2018

On the Schrödinger-Debye System in Compact Riemannian Manifolds

We consider the initial value problem (IVP) associated to the Schrödinger-Debye system posed on a $d$-dimensional compact Riemannian manifold $M$ and prove local well-posedness result for given data $(u_0, v_0)\in H^s(M)\times (H^s(M)\cap L^{\infty}(M))$ whenever $s>\frac{d}2-\frac12$, $d\geq 2$. For $d=2$, we apply a sharp version of the Gagliardo-Nirenberg inequality in compact manifold to derive an a priori estimate for the $H^1$-solution and use it to prove the global well-posedness result in this space.

math.AP↗