arXiv:2510.06188v2 Announce Type: replace 
Abstract: Real-time speech assistants are becoming increasingly popular for ensuring improved accessibility to information. Bengali, being a low-resource language with a high regional dialectal diversity, has seen limited progress in developing such systems. Existing systems are not optimized for real-time use and focus only on standard Bengali. In this work, we present BanglaTalk, the first real-time speech assistance system for Bengali regional dialects. BanglaTalk follows the client-server architecture and uses the Real-time Transport Protocol (RTP) to ensure low-latency communication. To address dialectal variation, we introduce a dialect-aware ASR system, BRDialect, developed by fine-tuning the IndicWav2Vec model in ten Bengali regional dialects. It outperforms the baseline ASR models by 12.41-33.98% on the RegSpeech12 dataset. Furthermore, BanglaTalk can operate at a low bandwidth of 24 kbps while maintaining an average end-to-end delay of 4.9 seconds. Low bandwidth usage and minimal end-to-end delay make the system both cost-effective and interactive for real-time use cases, enabling inclusive and accessible speech technology for the diverse community of Bengali speakers. Code is available in https://github.com/Jak57/BanglaTalk

بانغلا توك هو نظام مبتكر للمساعدة الصوتية في الوقت الحقيقي مصمم خصيصًا للهجات الإقليمية البنغالية، حيث يتناول التحديات التي يطرحها وضع اللغة كموارد منخفضة وتنوعها اللهجي. على عكس الأنظمة الحالية التي تركز فقط على البنغالية القياسية، يستخدم بانغلا توك بنية عميل-خادم ويستفيد من بروتوكول النقل في الوقت الحقيقي (RTP) لضمان اتصال منخفض الكمون. يتضمن النظام مكونًا للتعرف على الكلام التلقائي (ASR) مدرك للهجات، BRDialect، الذي أظهر تحسينات ملحوظة مقارنة بالنماذج الأساسية على مجموعة بيانات RegSpeech12.

BanglaTalk es un sistema pionero de asistencia de voz en tiempo real diseñado específicamente para los dialectos regionales bengalíes, abordando los desafíos que plantea el estatus de baja disponibilidad de recursos del idioma y su diversidad dialectal. A diferencia de los sistemas existentes que se centran únicamente en el bengalí estándar, BanglaTalk emplea una arquitectura cliente-servidor y utiliza el Protocolo de Transporte en Tiempo Real (RTP) para garantizar una comunicación de baja latencia. El sistema incorpora un componente de Reconocimiento Automático de Voz (ASR) consciente de los …

BanglaTalk est un système d'assistance vocale en temps réel innovant, conçu spécifiquement pour les dialectes régionaux bengalis, répondant aux défis posés par le statut de langue à faibles ressources et la diversité dialectale. Contrairement aux systèmes existants qui se concentrent uniquement sur le bengali standard, BanglaTalk utilise une architecture client-serveur et le protocole de transport en temps réel (RTP) pour une communication à faible latence. Le système intègre un composant de reconnaissance vocale automatique (ASR) conscient des dialectes, BRDialect, qui a montré des améliorati…

BanglaTalk is a pioneering real-time speech assistance system designed specifically for Bengali regional dialects, addressing the challenges posed by the language's low-resource status and diverse dialectal variations. Unlike existing systems that focus solely on standard Bengali, BanglaTalk employs a client-server architecture and utilizes the Real-time Transport Protocol (RTP) for low-latency communication. The system incorporates a dialect-aware Automatic Speech Recognition (ASR) component, BRDialect, which has shown significant performance improvements over baseline models on the RegSpeech…

BanglaTalk: Towards Real-Time Speech Assistance for Bengali Regional Dialects

Here's everything you need to know about the latest smart ring by Oura, based on a wearable expert's real-world usage.

Is the $500 Oura Ring 4 Ceramic worth it? I wore one for a month, and here's my advice

Samsung's QN90F QLED delivers great streaming and gaming performance, making it a strong holiday value.

Why I recommend this Samsung QLED TV over pricier OLED models in 2025 - and don't regret it

<h1>
 
 
 AI Amnesia: Erasing Knowledge Without a Trace
</h1>

Imagine your AI model accidentally learned something it shouldn't have – sensitive customer data, for example. Current methods for deleting this information often require retraining the entire model, an expensive and time-consuming process. What if we could surgically remove that knowledge without starting from scratch?

The key lies in a novel approach: creating artificial "forgetting cues." We're teaching the model to unlearn specific data patterns by exposing it to carefully crafted synthetic examples. These examples are designed to strongly contradict the information we want the model to forget, effectively overwriting the problematic associations in its memory. This works even if you don’t have access to the original data you need to erase.

Think of it like this: you're trying to forget a bad song stuck in your head. Instead of trying to actively suppress it (which rarely works), you blast an even more catchy song. The new song overwrites the old one, effectively erasing it from your mental playlist.

Benefits of Selective Forgetting:

<ul>
<li> Enhanced Data Privacy: Remove sensitive data without compromising the overall model's performance.</li>
<li> Reduced Retraining Costs: Avoid full model retraining, saving significant time and resources.</li>
<li> Improved Model Security: Eliminate vulnerabilities introduced by unintentionally learned patterns.</li>
<li> Adaptable Learning: Enables continuous refinement of AI models based on evolving data landscapes.</li>
<li> Compliance Ready: Supports compliance with data privacy regulations like GDPR.</li>
<li> Scalable Solutions: Works efficiently even with limited access to training data.</li>
</ul>

Practical Tip: One challenge is ensuring the synthetic data accurately targets the information you want to remove without negatively impacting the model's ability to generalize. Rigorous testing and validation with holdout datasets are crucial.

The promise of AI that can truly 'forget' opens exciting possibilities for responsible AI development. By enabling precise data deletion, we pave the way for more secure, compliant, and adaptable machine learning systems. Imagine AI models that can adapt to changing ethical guidelines or quickly unlearn incorrect information, all without massive retraining efforts. This is a crucial step towards trustworthy and responsible AI that respects data privacy and aligns with societal values. Future exploration could include extending this to different data modalities and model architectures.

Related Keywords: machine unlearning, data privacy, few-shot learning, zero-shot learning, synthetic data, model editing, catastrophic forgetting, incremental learning, continual learning, deep learning, neural networks, data security, algorithmic fairness, responsible ai, ethical ai, federated unlearning, privacy-preserving ai, model retraining, AI governance, data deletion, GDPR compliance

يتناول المقال نهجًا جديدًا لمعالجة مشكلة نسيان الذكاء الاصطناعي، حيث قد تحتفظ نماذج الذكاء الاصطناعي عن غير قصد بمعلومات حساسة. تتطلب الطرق التقليدية لحذف هذه البيانات غالبًا إعادة تدريب كاملة للنموذج، وهو ما يعد مكلفًا ويستغرق وقتًا. تتضمن التقنية الجديدة إنشاء 'إشارات نسيان' اصطناعية تساعد النموذج على نسيان أنماط بيانات معينة من خلال تعريضه لأمثلة صناعية تتناقض مع المعلومات غير المرغوب فيها، مما يسمح بإزالة المعرفة المستهدفة دون الحاجة للوصول إلى البيانات الأصلية.

El artículo aborda un nuevo enfoque para tratar el problema de la amnesia de la IA, donde los modelos de IA pueden retener inadvertidamente información sensible. Los métodos tradicionales para eliminar estos datos a menudo requieren un reentrenamiento completo del modelo, lo que es costoso y lleva tiempo. La nueva técnica implica crear 'señales de olvido' artificiales que ayudan al modelo a desaprender patrones de datos específicos al exponerlo a ejemplos sintéticos que contradicen la información no deseada, permitiendo así la eliminación selectiva de conocimientos sin necesidad de acceder a l…

L'article traite d'une nouvelle approche pour résoudre le problème de l'amnésie de l'IA, où les modèles d'IA peuvent involontairement conserver des informations sensibles. Les méthodes traditionnelles pour supprimer ces données nécessitent souvent un réentraînement complet du modèle, ce qui est coûteux et long. La nouvelle technique consiste à créer des 'cues d'oubli' artificiels qui aident le modèle à désapprendre des schémas de données spécifiques en lui présentant des exemples synthétiques qui contredisent les informations indésirables, permettant ainsi une suppression ciblée des connaissan…

The article discusses a novel approach to address the issue of AI amnesia, where AI models may inadvertently retain sensitive information. Traditional methods for deleting such data often require complete retraining of the model, which is costly and time-consuming. The new technique involves creating artificial 'forgetting cues' that help the model unlearn specific data patterns by presenting it with synthetic examples that contradict the unwanted information, allowing for targeted knowledge removal without needing access to the original data.

AI Amnesia: Erasing Knowledge Without a Trace

Baseus' Enercore CG11 travel adapter is one of the better designed ones I've tested, although its best feature isn't immediately apparent.

Traveling soon? Why this one charger is the only one you'll ever need to pack

Turning materials like wood chips, crop residues and municipal solid waste into fuels and chemicals is important for our country's energy independence.

طور الباحثون نماذج حاسوبية متقدمة تهدف إلى تحسين التنبؤات لعمليات طحن الكتلة الحيوية. تركز هذه الابتكارات على تحويل مواد مثل رقائق الخشب، بقايا المحاصيل، والنفايات الصلبة البلدية إلى وقود ومواد كيميائية قيمة، وهو أمر مهم لتعزيز استقلالية الطاقة في البلاد. من المتوقع أن تعمل النماذج على تحسين كفاءة معالجة الكتلة الحيوية، مما يساهم في حلول الطاقة المستدامة.

Investigadores han desarrollado modelos informáticos avanzados destinados a mejorar las predicciones para los procesos de molienda de biomasa. Esta innovación se centra en convertir materiales como astillas de madera, residuos de cultivos y residuos sólidos municipales en combustibles y productos químicos valiosos, lo que es crucial para mejorar la independencia energética del país. Se espera que los modelos optimicen la eficiencia del procesamiento de biomasa, contribuyendo así a soluciones energéticas sostenibles.

Des chercheurs ont développé des modèles informatiques avancés visant à améliorer les prévisions pour les processus de broyage de la biomasse. Cette innovation se concentre sur la conversion de matériaux tels que les copeaux de bois, les résidus de culture et les déchets solides municipaux en combustibles et produits chimiques précieux, ce qui est crucial pour renforcer l'indépendance énergétique du pays. Les modèles devraient optimiser l'efficacité du traitement de la biomasse, contribuant ainsi à des solutions énergétiques durables.

Researchers have developed advanced computer models aimed at improving predictions for biomass milling processes. This innovation focuses on converting materials such as wood chips, crop residues, and municipal solid waste into valuable fuels and chemicals, which is crucial for enhancing energy independence in the country. The models are expected to optimize the efficiency of biomass processing, thereby contributing to sustainable energy solutions.

Researchers develop computer models for better biomass milling predictions

The most interesting aspect of the latest Even Realities glasses may be their limitations.

I've tried several AI smart glasses (including Meta Ray-Bans) in 2025 - these are the most comfortable

arXiv:2501.07251v3 Announce Type: replace 
Abstract: Crafting adversarial examples is crucial for evaluating and enhancing the robustness of Deep Neural Networks (DNNs), presenting a challenge equivalent to maximizing a non-differentiable 0-1 loss function.
  However, existing single objective methods, namely adversarial attacks focus on a surrogate loss function, do not fully harness the benefits of engaging multiple loss functions, as a result of insufficient understanding of their synergistic and conflicting nature.
  To overcome these limitations, we propose the Multi-Objective Set-based Attack (MOS Attack), a novel adversarial attack framework leveraging multiple loss functions and automatically uncovering their interrelations.
  The MOS Attack adopts a set-based multi-objective optimization strategy, enabling the incorporation of numerous loss functions without additional parameters.
  It also automatically mines synergistic patterns among various losses, facilitating the generation of potent adversarial attacks with fewer objectives.
  Extensive experiments have shown that our MOS Attack outperforms single-objective attacks. Furthermore, by harnessing the identified synergistic patterns, MOS Attack continues to show superior results with a reduced number of loss functions. Our code is available at https://github.com/pgg3/MOS-Attack.

يقدم إطار عمل MOS Attack نهجًا قابلًا للتوسع لتوليد أمثلة عدائية لشبكات الأعصاب العميقة (DNN). يتناول القيود المفروضة على الهجمات العدائية ذات الهدف الواحد من خلال الاستفادة من وظائف خسارة متعددة وعلاقاتها المتبادلة. تتيح هذه الاستراتيجية متعددة الأهداف دمج وظائف خسارة متنوعة دون الحاجة إلى معلمات إضافية، مما يعزز تقييم متانة شبكات الأعصاب العميقة ضد الهجمات العدائية.

El marco MOS Attack introduce un enfoque escalable para generar ejemplos adversariales para Redes Neuronales Profundas (DNN). Aborda las limitaciones de los ataques adversariales de objetivo único existentes al aprovechar múltiples funciones de pérdida y sus interrelaciones. Esta estrategia de optimización multiobjetivo permite la incorporación de varias funciones de pérdida sin necesidad de parámetros adicionales, mejorando así la evaluación de la robustez de las DNN frente a ataques adversariales.

Le cadre MOS Attack introduit une approche évolutive pour générer des exemples adversariaux pour les réseaux de neurones profonds (DNN). Il aborde les limitations des attaques adversariales à objectif unique existantes en exploitant plusieurs fonctions de perte et leurs interrelations. Cette stratégie d'optimisation multi-objectifs permet d'incorporer diverses fonctions de perte sans nécessiter de paramètres supplémentaires, améliorant ainsi l'évaluation de la robustesse des DNN face aux attaques adversariales.

The MOS Attack framework introduces a scalable approach to generating adversarial examples for Deep Neural Networks (DNNs). It addresses the limitations of existing single-objective adversarial attacks by leveraging multiple loss functions and their interrelations. This multi-objective optimization strategy allows for the incorporation of various loss functions without the need for additional parameters, enhancing the robustness evaluation of DNNs against adversarial attacks.

MOS-Attack: A Scalable Multi-objective Adversarial Attack Framework

arXiv:1910.03867v3 Announce Type: replace-cross 
Abstract: We present multi-point optimization: an optimization technique that allows to train several models simultaneously without the need to keep the parameters of each one individually. The proposed method is used for a thorough empirical analysis of the loss landscape of neural networks. By extensive experiments on FashionMNIST and CIFAR10 datasets we demonstrate two things: 1) loss surface is surprisingly diverse and intricate in terms of landscape patterns it contains, and 2) adding batch normalization makes it more smooth. Source code to reproduce all the reported results is available on GitHub: https://github.com/universome/loss-patterns.

يقدم المقال تقنية تحسين متعددة النقاط، وهي تقنية تسمح بتدريب عدة نماذج في وقت واحد دون الحاجة إلى إدارة معلمات كل نموذج بشكل فردي. تُستخدم هذه الطريقة لإجراء تحليل تجريبي شامل لمنظر خسارة الشبكات العصبية. من خلال تجارب واسعة على مجموعات بيانات FashionMNIST وCIFAR10، نوضح أمرين: 1) سطح الخسارة متنوع ومعقد بشكل مدهش من حيث الأنماط الموجودة فيه، و2) إضافة التطبيع بالدفعات تجعل السطح أكثر سلاسة. الشيفرة المصدرية لإعادة إنتاج جميع النتائج المبلغ عنها متاحة على GitHub.

El artículo presenta la optimización multipunto, una técnica que permite entrenar varios modelos simultáneamente sin necesidad de gestionar individualmente sus parámetros. Este método facilita un análisis empírico del paisaje de pérdidas de las redes neuronales. Los experimentos realizados en los conjuntos de datos FashionMNIST y CIFAR10 revelan que la superficie de pérdidas es compleja y diversa, mientras que la incorporación de la normalización por lotes contribuye a un paisaje de pérdidas más suave. El código fuente para reproducir los resultados está disponible en GitHub.

L'article présente l'optimisation multi-point, une technique permettant de former plusieurs modèles simultanément sans gérer individuellement leurs paramètres. Cette méthode facilite une analyse empirique du paysage de perte des réseaux de neurones. Les expériences menées sur les ensembles de données FashionMNIST et CIFAR10 révèlent que la surface de perte est complexe et diversifiée, tandis que l'incorporation de la normalisation par lot contribue à un paysage de perte plus lisse. Le code source pour reproduire les résultats est disponible sur GitHub.

The article presents multi-point optimization, a technique for training multiple models simultaneously without individually managing their parameters. This method facilitates an empirical analysis of the loss landscape of neural networks. Experiments conducted on FashionMNIST and CIFAR10 datasets reveal that the loss surface is complex and diverse, while the incorporation of batch normalization contributes to a smoother loss landscape. The source code for replicating the results is available on GitHub.

Loss Patterns of Neural Networks

arXiv:2510.24021v2 Announce Type: replace 
Abstract: Knowledge distillation (KD) is a standard route to compress Large Language Models (LLMs) into compact students, yet most pipelines uniformly apply token-wise loss regardless of teacher confidence. This indiscriminate supervision amplifies noisy, high-entropy signals and is especially harmful under large teacher-student capacity gaps. We introduce SelecTKD, a plug-and-play Selective Token-Weighted distillation framework that shifts the focus from "how to measure divergence" to "where to apply learning". At each step, the student proposes tokens that are verified by the teacher through a robust propose-and-verify procedure with two variants: greedy Top-k and non-greedy Spec-k. Accepted tokens receive full loss, while rejected tokens are masked or down-weighted. This objective-agnostic design works with on- and off-policy data, induces an implicit curriculum quantified by Token Acceptance Rate (TAR), and stabilizes optimization. Across instruction following, mathematical reasoning, code generation, and a VLM setting, SelecTKD consistently improves strong baselines and achieves state-of-the-art results for small models without architectural changes or extra reference models.

SelecTKD: Selective Token-Weighted Knowledge Distillation for LLMs

arXiv:2511.13368v1 Announce Type: new 
Abstract: Large language models (LLMs) perform strongly across tasks and languages, yet how improvements in one task or language affect other tasks and languages and their combinations remains poorly understood. We conduct a controlled PEFT/LoRA study across multiple open-weight LLM families and sizes, treating task and language as transfer axes while conditioning on model family and size; we fine-tune each model on a single task-language source and measure transfer as the percentage-point change versus its baseline score when evaluated on all other task-language target pairs. We decompose transfer into (i) Matched-Task (Cross-Language), (ii) Matched-Language (Cross-Task), and (iii) Cross-Task (Cross-Language) regimes. We uncover two consistent general patterns. First, a pronounced on-task vs. off-task asymmetry: Matched-Task (Cross-Language) transfer is reliably positive, whereas off-task transfer often incurs collateral degradation. Second, a stable donor-recipient structure across languages and tasks (hub donors vs. brittle recipients). We outline implications for risk-aware fine-tuning and model specialisation.

Donors and Recipients: On Asymmetric Transfer Across Tasks and Languages with Parameter-Efficient Fine-Tuning

arXiv:2511.11878v1 Announce Type: new 
Abstract: While large language models (LLMs) show transformative potential in healthcare, their development remains focused on high-resource languages, creating a critical barrier for others as simple translation fails to capture unique clinical and cultural nuances, such as endemic diseases. To address this, we introduce MedPT, the first large-scale, real-world corpus for Brazilian Portuguese, comprising 384,095 authentic question-answer pairs from patient-doctor interactions. The dataset underwent a meticulous multi-stage curation protocol, using a hybrid quantitative-qualitative analysis to filter noise and contextually enrich thousands of ambiguous queries. We further augmented the corpus via LLM-driven annotation, classifying questions into seven semantic types to capture user intent. Our analysis reveals its thematic breadth (3,200 topics) and unique linguistic properties, like the natural asymmetry in patient-doctor communication. To validate its utility, we benchmark a medical specialty routing task: fine-tuning a 1.7B parameter model achieves an outstanding 94\% F1-score on a 20-class setup. Furthermore, our qualitative error analysis shows misclassifications are not random but reflect genuine clinical ambiguities (e.g., between comorbid conditions), proving the dataset's deep semantic richness. We publicly release MedPT to foster the development of more equitable, accurate, and culturally-aware medical technologies for the Portuguese-speaking world.

BanglaTalk: Towards Real-Time Speech Assistance for Bengali Regional Dialects

Was this article worth reading? Share it