AI & ML

A focused conversational assistant (FCA) realizes human-computer interaction bounded in a predefined scope of operation. With the advent of large language models (LLMs), it has become imperative to integrate them in conversational assistants (CAs). However, an LLM can become largely inaccurate in an FCA with multiple responsibilities, like information extraction, scope adherence and response generation. In this paper, we attempt to use an LLM for an FCA while constricting the scope of operation and maintaining a guided flow of conversation. We present a strategical combination of discriminative AI methods and generative AI models. Our methodology includes (i) a component of natural language understanding (NLU) operating discriminatively, (ii) a conditional intent-based routing of user messages to appropriate response generators, and (iii) response generators which are either custom ones or open sourced LLMs. The collation of these three strategies realizes a hybrid AI system, assisting FCA with adhering to the defined scope, maintaining context and dialogue flow.

Ensuring auditability and verifiability in Federated Learning (FL) is both challenging and essential to guarantee that local data remains untampered and client updates are trustworthy. Recent FL frameworks assess client contributions through a trusted central server using various client selection and aggregation techniques. However, reliance on a central server can create a single point of failure, making it vulnerable to privacy-centric attacks and limiting its ability to audit and verify client-side data contributions due to restricted access. In addition, data quality and fairness evaluations are often inadequate, failing to distinguish between high-impact contributions and those from low-quality or poisoned data. To address these challenges, we propose Federated Auditable and Verifiable Data valuation (FAVD), a privacy-preserving method that ensures auditability and verifiability of client contributions through data valuation, independent of any central authority or predefined training algorithm. FAVD utilizes shared local data density functions to construct a global density function, aligning data contributions and facilitating effective valuation prior to local model training. This proactive approach improves transparency in data valuation and ensures that only benign updates are generated, even in the presence of malicious data. Further, to mitigate privacy risks associated with sharing data density functions, we add Gaussian noise to each client’s local density function before sharing it with the server. We theoretically demonstrate the convergence, auditability, and verifiability of FAVD, along with its resilience against data poisoning threats. Our experiments on five diverse benchmarks, including three medical datasets, show that FAVD achieves significant performance gains, accurate data valuation, and fair client contributions under threat, highlighting its reliability as a trustworthy FL approach.

Anomaly detection is needed in many applications, such as video surveillance, manufacturing defect detection, and medical image analysis. However, this poses a significant challenge as anomalies can exhibit diverse patterns. Furthermore, the scarcity of labelled anomaly (defective) images complicates the use of supervised classification approaches. To address this, alternative approaches such as unsupervised or self-supervised learning are explored, effectively overcoming challenges posed by limited labelled data. While several techniques exist, there is still significant room for improvement in overall detection results, especially in handling noisy images, which is a realistic dimension often overlooked in existing methods. Therefore, we have proposed a novel framework consisting of various modules, including the Attention-Aware PDE constraint Feature Denoiser Module, Teacher, Student, and Anomaly-Attenuator. Here, we have utilized the loss function in a novel manner, which improves overall performance and ensures consistency. Additionally, we train the network in such a way that a single model would work for multiple types of objects. Finally, we considered different datasets for testing and observed that our model provides superior results, outperforming existing state-of-the-art methods for noisy and clean images.

Intent recognition aids conversational AI in understanding the meaning of user messages, which has to be realised in multiple languages for multilingual scenarios. In this paper, we explore the effectiveness and limitations of massively multilingual language representational models like mBERT and XLM-R, in expressing their crosslingual transfer learning ability for multilingual intent recognition. It is noted that models based on BERT can be fine-tuned for multiple downstream tasks with a limited domain-specific text dataset, in order to achieve task and domain adaptation. We analyse the particulars of text data that assist these models in effectively expressing crosslingual transfer, without an explicit fine-tuning. We noticed that when the text content across multiple languages is “inherently parallel fine-tuning with domain-specific text from one language successfully exhibit crosslingual transfer to other languages for recognising intents from short text messages. Inherently parallel text data has content in multiple languages with similar meaning, though not in parallel sentences. Such data can be found commonly in messages received by conversational assistants, in the form of customer’s queries, reviews, or requests related to same products or services offered by companies/providers. The analysis presented in this paper can benefit numerous business and industrial applications in rich multilingual societies, where it is easier to collect and prepare text data for fine-tuning in one major language, and build conversational assistants catering to multiple target low-resource languages.