Search

Optimizing a transmit multiple-input–multiple-output (MIMO) radar waveform subject to the nonconvex constant modulus constraint (CMC) remains a problem of enduring interest. The past decade has seen a variety of tailored iterative approaches with various performance–complexity tradeoffs. Despite promising work, iterative algorithms have a speed handicap and require meticulous parameter tuning. Deep learning has recently been proposed for MIMO radar waveform design, where iterative optimization is replaced by a neural network that regresses from the problem specification to the desired waveform. Once trained, a deep network can quickly regress the desired waveform coefficients, but it is a black box and may excel only when generous training is available. We present a fast, learned, and explainable deep (FLED) learning approach by unrolling a state-of-the-art iterative algorithm. We particularly leverage the recently proposed projection–descent–retraction (PDR) algorithm and design a neural network where each PDR iteration is mapped to a step of the network while preserving the CMC. Alongside an interpretable network architecture, we propose learning of key algorithm parameters such as the direction of descent, while introducing learning of FLED initialization and solution refinement through trainable prune–expand blocks. By enforcing appropriate regularizers in the learning of these parameters, FLED breaks the tradeoff between computational speed and performance. Compared to the state-of-the-art alternatives, it is near real time with boosted performance—fidelity to the desired beampattern. Crucially, the benefits of FLED over competing deep learning techniques are enhanced in the regime of limited training, i.e., FLED exhibits superior generalizability.

Graphical Abstract:

Amidst the rapid dissemination of news, preventing misleading information is crucial. Using sentiment score, count vectorization, Term frequency and inverse document frequency, this study presents an innovative way for recognizing misleading information. Across three datasets eg. Covid-19, LIAR and Politifact, the proposed method evaluated accuracy and F1-score using different machine learning techniques. For 2-class classification on the LIAR dataset, the model achieves a 20% performance boost compared to state-of-the-art deep learning and machine techniques, and shows a significant 30% improvement for 6-class classification. In addition, compared to deep learning methods, machine learning approaches provide better accuracy and F1-score while requiring less time to complete. This study offers a potential strategy to tackle the spread of false news in the modern information ecosystem.

Wireless communication systems are inherently challenged by factors such as fading, path loss, and shadowing, leading to potential errors in data transmission. Traditional methods to mitigate these issues include power control, diversification, variable beamforming, and modulation techniques. However, the unpredictable nature of the wireless medium often limits their effectiveness. A new approach to address these challenges is the implementation of cascaded intelligent reflecting surfaces (IRS). IRS systems consist of multiple passive elements that intelligently reflect electromagnetic waves, thereby enhancing signal quality. The Advanced Discrete Fourier Transform (ADFT) matrix scheme is explored in channel estimation, a novel method particularly suitable for wireless networks utilising cascaded IRS. The ADFT matrix scheme is significant for its efficiency in managing the common‐link configuration of cascading channel coefficients, which effectively reduces pilot overhead. When compared to traditional channel estimation methods like the Least Square|least squares, Maximal a posteriori probability, and Linear Minimum Mean Square Error, the ADFT matrix scheme exhibits superior performance. It achieves a remarkable reduction in normalised mean squared error (NMSE) – 66% and 80% at 20 dB and 15 dB Signal to‐Noise Ratios (SNR), respectively. Furthermore, increasing the pilot length correlates with enhanced NMSE performance, with a noted 33% improvement as the base station distance increases. Simulations demonstrate that with an escalation in the number of IRS elements and SNR, the ADFT matrix scheme consistently surpasses conventional methods. This advancement represents a significant leap in the field of wireless communication technology. An advanced Cascaded Intelligent Reflecting Surfaces (IRS) system enriched with passive elements for enhanced wireless communication optimisation.

Near-infrared (NIR) emitting materials find a variety of applications in photodynamic therapy, national security, food inspection, and bioimaging, but achieving efficient NIR luminescence in lead-free perovskite materials is still challenging. Herein, we synthesized a series of lead-free Mo4+alloyed Cs2ZrCl6vacancy-ordered double perovskites with different Mo concentrations. These materials display tunable dual luminescence bands due to intrinsic host emission in the blue region and dopant induced d–d transitions emitting in the NIR region. Host- and dopant-induced emission can be excited at different energies, as revealed by spectroscopic studies. Minimum interaction between the host and dopant octahedra accounts for the dual emission in these 0D materials, further confirmed by density functional theory (DFT) calculations. Alloying Mo4+reduces the band gap of these materials and DFT calculations confirm that the decrease in band gap is due to the increased contribution of Mo 4d states at the conduction band minima. Moreover, the structural stability of the host material also increases after Mo4+alloying. Our study explored the tunable dual band emission in visible and NIR regions and enhanced structural stability in Mo4+alloyed Cs2ZrCl6, thereby opening new avenues in the development of Pb-free NIR-emitting phosphors.

Type 2 diabetes mellitus is a major risk factor for hepatocellular carcinoma (HCC). Changes in extracellular matrix (ECM) mechanics contribute to cancer development1,2, and increased stiffness is known to promote HCC progression in cirrhotic conditions3,4. Type 2 diabetes mellitus is characterized by an accumulation of advanced glycation end-products (AGEs) in the ECM; however, how this affects HCC in non-cirrhotic conditions is unclear. Here we find that, in patients and animal models, AGEs promote changes in collagen architecture and enhance ECM viscoelasticity, with greater viscous dissipation and faster stress relaxation, but not changes in stiffness. High AGEs and viscoelasticity combined with oncogenic β-catenin signalling promote HCC induction, whereas inhibiting AGE production, reconstituting the AGE clearance receptor AGER1 or breaking AGE-mediated collagen cross-links reduces viscoelasticity and HCC growth. Matrix analysis and computational modelling demonstrate that lower interconnectivity of AGE-bundled collagen matrix, marked by shorter fibre length and greater heterogeneity, enhances viscoelasticity. Mechanistically, animal studies and 3D cell cultures show that enhanced viscoelasticity promotes HCC cell proliferation and invasion through an integrin-β1–tensin-1–YAP mechanotransductive pathway. These results reveal that AGE-mediated structural changes enhance ECM viscoelasticity, and that viscoelasticity can promote cancer progression in vivo, independent of stiffness.

Hepatocellular carcinoma (HCC) represents a global health challenge, and ranks among one of the most prevalent and deadliest cancers worldwide. Therapeutic advances have expanded the treatment armamentarium for patients with advanced HCC, but obstacles remain. Precision oncology, which aims to match specific therapies to patients who have tumours with particular features, holds great promise. However, its implementation has been hindered by the existence of numerous ‘HCC influencers’ that contribute to the high inter-patient heterogeneity. HCC influencers include tumour-related characteristics, such as genetic alterations, immune infiltration, stromal composition and aetiology, and patient-specific factors, such as sex, age, germline variants and the microbiome. This Review delves into the intricate world of HCC, describing the most innovative model systems that can be harnessed to identify precision and/or personalized therapies. We provide examples of how different models have been used to nominate candidate biomarkers, their limitations and strategies to optimize such models. We also highlight the importance of reproducing distinct HCC influencers in a flexible and modular way, with the aim of dissecting their relative contribution to therapy response. Next-generation HCC models will pave the way for faster discovery of precision therapies for patients with advanced HCC.

Beyond achievement, educational settings offer informal supports that may be critical for child and youth mental health. However, children’s educational environments have experienced significant disruption with the coronavirus pandemic. School settings offer unique opportunities to support children’s mental health, but research must identify powerful points of intervention. This study examined school factors (aspirations, perceived competence, sense of belonging, and emotional engagement) as predictors of children’s mental health, and the potential consequences of increasing screen time in and outside of school. Participants (N= 707) were parents and their children (6–18 years) from community and clinical settings who completed prospective surveys about children’s school experiences and mental health symptoms (November 2020-May 2022). Standardized measures of depression, anxiety, irritability, inattention, and hyperactivity were collected. Structural equation modelling tested longitudinal associations between screen time, school factors, and mental health outcomes. Positive associations between each of the school factors (B =0.14 [SE = 0.04] to B= 0.43 [SE = 0.04]) suggested they may reinforce one another. Longitudinally, sense of belonging and emotional engagement at school predicted lower severity for symptoms of depression, anxiety, irritability, and inattention (B=-0.14 [SE = 0.07] to B =-0.33 [SE = 0.10]). Greater screen time was associated with lower aspirations and perceived competence (B = −0.08 [SE = 0.04] to B = −0.13 [SE = 0.06]). Results suggest that school factors beyond achievement may be key correlates of child and youth mental health. While curriculum expectations emphasize academic achievement, an investment in supporting positive attitudes and aspirations at school is also warranted.

Twelve 3,5-disubstituted-thiazolidine-2,4-dione (TZD) hybrids were synthesized using solution phase chemistry. Continuing our previous work, nine O-modified ethyl vanillin (8a–i) derivatives were synthesized and reacted with the TZD core viaKnoevenagel condensation under primary reaction conditions to obtain final derivatives 9a–i. Additionally, three isatin–TZD hybrids (11a–c) were synthesized. The intermediates and final derivatives were characterized using 1H and 13C NMR spectroscopy, and the observed chemical shifts agreed with the proposed structures. The in vitroalpha-amylase and alpha-glucosidase inhibitory evaluation of newly synthesized derivatives revealed compounds 9Fand 9Gas the best dual inhibitors, with IC50values of 9.8 ± 0.047 μM for alpha-glucosidase (9F) and 5.15 ± 0.0017 μM for alpha-glucosidase (9G), 17.10 ± 0.015 μM for alpha-amylase (9F), and 9.2 ± 0.092 μM for alpha-amylase (9G). The docking analysis of synthesized compounds indicated that compounds have a higher binding affinity for alpha-glucosidase as compared to alpha-amylase, as seen from docking scores ranging from −1.202 to −5.467 (for alpha-amylase) and −4.373 to −7.300 (for alpha-glucosidase). Further, the molecules possess a high LD50value, typically ranging from 1000 to 1600 mg kg−1of body weight, and exhibit non-toxic properties. The in vitrocytotoxicity assay results on PANC-1 and INS-1 cells demonstrated that the compounds were devoid of significant toxicity against the tested cells. Compounds 9Fand 9Gshowed high oral absorption, i.e., oral absorption >96%, and their molecular dynamics simulation yielded results closely aligned with the observed docking outcomes. Finally, compounds 9Fand 9Gwere evaluated for in vivoantidiabetic assessment by the induction of diabetes in Wistar rats using streptozotocin. Molecule 9Ghas been identified as the most effective anti-diabetic molecule due to its ability to modulate several biochemical markers in blood plasma and tissue homogenates. The results were further confirmed by histology investigations conducted on isolated pancreas, liver, and kidney.

Background: Follicular lymphoma (FL) and marginal zone lymphoma (MZL) are types of indolent non-Hodgkin lymphoma (NHL) that develop in the B lymphocytes (also known as B cells). Objective: The aim of this study was to conduct a comprehensive review of studies relating to cost effectiveness, costs and resource use, and health-related quality of life (HRQoL) in patients with FL or MZL. Methods: Three separate systematic reviews were conducted to identify all published evidence on cost effectiveness, costs and resource use, and HRQoL between 2007 and March 2017 using the MEDLINE®, MEDLINE in-process, E-pubs ahead of print (Ovid SP®), Embase (Ovid SP®), NHS EED, and EconLit databases. Select congress proceedings were also searched. Two systematic reviewers independently reviewed titles, abstracts, and full papers against eligibility criteria. Relevant data were extracted into bespoke data extraction templates (DETs) by a single systematic reviewer; these data were then validated for accuracy by a second reviewer against clean copies of the relevant publications. Results: A total of 25 cost-effectiveness studies (24 in FL; 1 in FL and MZL) met the eligibility criteria. Markov models were the most utilised cost-effectiveness model. US FL studies reported an incremental cost-effectiveness ratio (ICER) of $28,565/QALY for first-line rituximab–cyclophosphamide, vincristine, and prednisone (R-CVP) versus CVP, and $43,000/QALY for second-line obinutuzumab plus bendamustine (G + B) followed by G maintenance versus B. In the UK, ICERs were £1529–10,834/quality-adjusted life-year (QALY) for first-line rituximab + chemotherapy versus chemotherapy, £27,988/QALY for second-line G + B + G-maintenance versus B, and £62,653/QALY for second-line idelalisib versus chemotherapy and/or rituximab. Five costs/resource use and four HRQoL studies were identified in FL, and none in MZL. US mean lifetime costs in first-line patients ranged from $108,000 (rituximab) to $130,300 (rituximab–cyclophosphamide, doxorubicin hydrochloride, vincristine and prednisolone [CHOP]), and from £2185 (watch-and-wait) to £17,054 (chemotherapy) in the UK. In a multinational study, more rituximab-refractory patients receiving G + B + G-maintenance reported a meaningful improvement in total FACT-Lym scores compared with patients receiving B. In the UK, total FACT-Lym scores were meaningfully higher for newly diagnosed patients compared with patients with progression (136.04 vs. 109.7). Conclusions and Relevance: We found a small body of evidence of quality of life, and potentially cost-effective treatment options for FL; however, no evidence was reported on MZL specifically. The significant data gaps in knowledge in these diseases demonstrate a marked need for further studies.

Objective: Organizational health equity capacity assessments (OCAs) provide a valuable starting point to understand and strengthen an organization’s readiness and capacity for health equity. We conducted a scoping review to identify and characterize existing OCAs.Methods: We searched the PubMed, Embase, and Cochrane databases and practitioner websites to identify peer-reviewed and gray literature articles and tools that measure or assess health equity–related capacity in public health organizations. Seventeen OCAs met the inclusion criteria. We organized primary OCA characteristics and implementation evidence and described them thematically according to key categories.Results: All identified OCAs assessed organizational readiness or capacity for health equity, and many aimed to guide health equity capacity development. The OCAs differed in regard to thematic focus, structure, and intended audience. Implementation evidence was limited.Conclusions: By providing a synthesis of OCAs, these findings can assist public health organizations in selecting and implementing OCAs to assess, strengthen, and monitor their internal organizational capacity for health equity. This synthesis also fills a knowledge gap for those who may be considering developing similar tools in the future.

As the demographic structure of the global population continues to shift toward an aging population, caregivers encounter formidable obstacles in detecting wetness events and assessing the quantity of voided volume within diapers. Nevertheless, the integration of electronics with diapers, frequent removal and reapplication, and battery management still pose practical challenges. In this study, a solution of carbon- and silver-printed flexible capacitive sensors interfaced with a custom-designed application-specific integrated circuit (ASIC) is proposed to detect urination events and estimate voided volumes in diapers. A power-optimized application-specific front-end interface circuit is developed to enhance power efficiency and reduce battery maintenance. The circuit is designed to support power-harvesting-based operations and to reduce reliance on external sources, consuming a maximum power of 3.05 $\mu \text{W}$ . In vitro measurements validate on-chip electronics functionality and sensor performance to detect and quantify voided fluid volume as low as 50 mL. The promising results pave the way for cost-effective, disposable solutions in smart diapers. The proposed solution improves efficiency and comfort for both caregivers and elderly individuals.

AbstractExcess of androgens leads to various diseases such as Poly-Cystic Ovarian Syndrome, Prostate Cancer, Hirsutism, Obesity and Acne. 17β-Hydroxysteroid Dehydrogenase type 5 (17β-HSD5) converts androstenedione into testosterone peripherally, thereby significantly contributing to the development of these diseases. Indole-bearing scaffolds are reported as potential 17β-HSD5 inhibitors for the manifestation of diseases arising due to androgen excess. In the present work, we have extensively performed a combination of molecular docking, Gaussian field-based 3D-QSAR, Pharmacophore mapping and MD-simulation studies (100 ns) to identify the pharmacophoric features of indole-based compounds as potent 17β-HSD5 inhibitors. Molecular simulation studies of the most potent compound in the binding pocket of enzyme revealed that the compound 11was stable in the binding pocket and showed good binding affinity through interactions with various residues of active site pocket. The Molecular mechanics Generalized Born surface area continuum solvation (MM/GBSA) and Molecular mechanics Poisson–Boltzmann surface area (MM/PBSA) calculations revealed that the compound 11possessed a free binding energy of −36.36 kcal/mol and −7.00 kcal/mol, respectively, which was better as compared to reference compound Desmethyl indomethacin (DES). The developed pharmacophore will be helpful to design novel indole-based molecules as potent 17β-HSD5 inhibitors for the treatment of various androgenic disorders.Communicated by Ramaswamy H. Sarma

In this paper we prove some results concerning the bound estimates of the eigenvalues of a matrix polynomial. This in particular include the results earlier proved by Dehmer and Killian [On bounds for the zeros of univariate polynomials,World Academy of Science, Engineering and Technology, 26 (2007)] for the location of zeros of polynomials.

ABSTRACTIntroductionThe quinolone scaffold is a bicyclic benzene-pyridinic ring scaffold with nitrogen at the first position and a carbonyl group at the second or fourth position. It is endowed with a diverse spectrum of pharmacological activities, including antitumor activity, and has progressed into various development phases of clinical trials for their target-specific anticancer activity.Areas coveredThe present review covers both classes of quinolones, i.e.quinolin-2(H)-one and quinolin-4(H)-one as anticancer agents, along with their possible mode of binding. Furthermore, their structure-activity relationships, molecular mechanisms, and pharmacokinetic properties are also covered to provide insight into their structural requirements for their rational design as anticancer agents.Expert opinionSynthetic feasibility and ease of derivatization at multiple positions, has allowed medicinal chemists to explore quinolones and their chemical diversity to discover newer anticancer agents. The presence of both hydrogen bond donor (−NH) and acceptor (-C=O) functionality in the basic scaffold at two different positions, has broadened the research scope. In particular, substitution at the -NH functionality of the quinolone motif has provided ample space for suitable functionalization and appropriate substitution at the quinolone’s third, sixth, and seventh carbons, resulting in selective anticancer agents binding specifically with various drug targets.

The pandemic of SARS-CoV-2 or COVID-19 has hugely created an economic imbalance worldwide. With the exponential increase in the number of cases and to keep in check on the community transmission, there is high demand and acute shortage of diagnostic kits. The pooled-sample strategy turns out to be the promising strategy intended to determine the optimal testing for specimens with limited resources and without losing the test sensitivity and specificity. The study was performed with standard molecular biology graded lab equipment, FDA-approved COVID-19 RNA extraction, and SARS-CoV-2 tests kits.

Although rare compared with adult liver cancers, hepatoblastoma (HB) is the most common pediatric liver malignancy, and its incidence is increasing. Currently, the treatment includes surgical resection with or without chemotherapy, and in severe cases, liver transplantation in children. The effort to develop more targeted, HB-specific therapies has been stymied by the lack of fundamental knowledge about HB biology. Heat shock factor 1 (HSF1), a transcription factor, is a canonical inducer of heat shock proteins, which act as chaperone proteins to prevent or undo protein misfolding. Recent work has shown a role for HSF1 in cancer beyond the canonical heat shock response. The current study found increased HSF1 signaling in HB versus normal liver. It showed that less differentiated, more embryonic tumors had higher levels of HSF1 than more differentiated, more fetal-appearing tumors. Most strikingly, HSF1 expression levels correlated with mortality. This study used a mouse model of HB to test the effect of inhibiting HSF1 early in tumor development on cancer growth. HSF1 inhibition resulted in fewer and smaller tumors, suggesting HSF1 is needed for aggressive tumor growth. Moreover, HSF1 inhibition also increased apoptosis in tumor foci. These data suggest that HSF1 may be a viable pharmacologic target for HB treatment.

This article has supplementary material on the web site: www.jdbp.org.

The value of involving people with lived experience in the research process (ie, patient engagement) is increasingly being recognised within youth mental health research. The wide-reaching benefits of patient engagement have been documented in the literature, including the empowerment of patients, improvement of research outcomes, and increased relevance of research findings. Although various models exist to guide patient engagement, there are key concepts that deserve exploration to ensure the authentic implementation of these models and development of patient roles. Our Personal View aims to: identify and discuss barriers to patient engagement roles in the context of youth mental health research; consider how key concepts of relational empowerment, fluidity, and flexibility can address some of these barriers; and provide tangible recommendations for implementing authentic patient engagement throughout the research process.

A sustainable high heat transfer coefficient is critical for water and energy systems that utilize steam condensation. Current techniques cannot achieve a high heat transfer performance at a high heat flux. It is desired to separate vapor and liquid flow to provide a large surface area for condensation. Here, we report sustainable phase separation on microchannel-elevated micromembrane (MEM) by (1) separating the vapor and liquid using a hydrophobic micromembrane that rapidly removes condensed droplets to achieve a high heat transfer coefficient and (2) sustaining continual condensate removal inside the hydrophobic microchannels beneath the micromembrane to prevent flooding. We found that MEM could sustain phase separation at a heat flux of 1,000 kW/m2without apparent failure. The heat transfer coefficient is 300% higher than that of the dropwise condensation on a hydrophobic flat surface. The phase separation on MEM provides a paradigm for sustainable high-performance condensation regardless of subcooling.

Deep learning has enabled significant improvements in semantic image segmentation, especially in underwater imaging domains such as side scan sonar (SSS). In this work, we apply deep learning to synthetic aperture sonar (SAS) imagery, which has an advantage over traditional SSS in which SAS produces coherent high- and constant-resolution imagery. Despite the successes of deep learning, one drawback is the need for abundant labeled training data to enable success. Such abundant labeled data are not always available as in the case of SAS where collections are expensive and obtaining quality ground-truth labels may require diver intervention. To overcome these challenges, we propose a domain-specific deep learning network architecture utilizing a unique property to complex-valued SAS imagery: the ability to resolve angle-of-arrival (AoA) of acoustic returns through $k$ -space processing. By sweeping through consecutive incrementally advanced AoA bandpass filters (a process sometimes referred to as multi-look processing), this technique generates a sequence of images emphasizing angle-dependent seafloor scattering and motion from biologics along the seafloor or in the water column. Our proposal, which we call multi-look sequence processing network (MLSP-Net), is a domain-enriched deep neural network architecture that models the multi-look image sequence using a recurrent neural network (RNN) to extract robust features suitable for semantic segmentation of the seafloor without the need for abundant training data. Unlike previous segmentation works in SAS, our model ingests a complex-valued SAS image and affords the ability to learn the AoA filters in $k$ -space as part of the training procedure. We show the results on a challenging real-world SAS database, and despite the lack of abundant training data, our proposed method shows superior results over state-of-the-art techniques.

Expressed on epidermal Langerhans cells, CD1a presents a range of self-lipid antigens found within the skin; however, the extent to which CD1a presents microbial ligands from bacteria colonizing the skin is unclear. Here we identified CD1a-dependent T cell responses to phosphatidylglycerol (PG), a ubiquitous bacterial membrane phospholipid, as well as to lysylPG, a modified PG, present in several Gram-positive bacteria and highly abundant in Staphylococcus aureus. The crystal structure of the CD1a–PG complex showed that the acyl chains were buried within the A′- and F′-pockets of CD1a, while the phosphoglycerol headgroup remained solvent exposed in the F′-portal and was available for T cell receptor contact. Using lysylPG and PG-loaded CD1a tetramers, we identified T cells in peripheral blood and in skin that respond to these lipids in a dose-dependent manner. Tetramer+CD4+T cell lines secreted type 2 helper T cell cytokines in response to phosphatidylglycerols as well as to co-cultures of CD1a+dendritic cells and Staphylococcusbacteria. The expansion in patients with atopic dermatitis of CD4+CD1a–(lysyl)PG tetramer+T cells suggests a response to lipids made by bacteria associated with atopic dermatitis and provides a link supporting involvement of PG-based lipid-activated T cells in atopic dermatitis pathogenesis.

Underwater sites affected by unexploded ordnance (UXO) pose significant risks to both human safety and environmental well-being. Sonar imaging is commonly employed to investigate such sites and aid in UXO remediation efforts. However, manually identifying and classifying potential targets in sonar data is challenging and time-consuming. Previous research has explored the use of machine-learning models to recognize and categorize targets; however, many of these approaches lack transparency and fail to consider the underlying physical acoustics. Additionally, acquiring sufficient training data for these models can often be problematic. In this study, we present a novel approach by designing neural networks that explicitly account for the unique physics involved in the problem domain. UXOs examined using low-frequency sound frequently exhibit resonant behavior, where the sound is reradiated after initial geometric scattering, owing to the elastic and compressional properties of the objects. Moreover, such resonant effects are typically absent in clutter objects, making them advantageous in discriminating UXOs from non-UXOs. Consequently, we propose several neural network architectures that leverage these resonant effects, utilizing 3D data obtained from a synthetic aperture sonar (SAS) imaging sonar. Our first proposal incorporates a recurrent neural network to model the physics-based correlation among adjacent time/spatial slices, originating from the resonant phenomena. For our second proposal, we employ intensity imagery of orthogonal projections of the 3D data cube, which capture shape-specific resonant scattering mechanisms unique to specific types of UXOs. To evaluate the effectiveness of our methods, we compare them against recent state-of-the-art (SOTA) algorithms using a real-world 3D SAS dataset. Remarkably, even when confronted with limited training data, our approaches consistently demonstrate superior results. Our findings highlight the significant potential of incorporating physical acoustics into neural network designs for UXO detection and classification, offering improved accuracy and efficiency in underwater remediation operations.