In the past few years, Language Models (LMs) have shown par-human capabilities in several domains. Despite their practical applications and exceeding user consumption, they are susceptible to jailbreaks when malicious input exploits the LM’s weaknesses, causing it to deviate from its intended behavior. Current defensive strategies either classify the input prompt as adversarial or prevent LMs from generating harmful outputs. However, it is challenging to explain the reason behind the malicious nature of the jailbreak, which results in a wide variety of closed-box approaches. In this research, we propose and demonstrate that system-prompt attention from Small Language Models (SLMs) can be used to characterize adversarial prompts, providing a novel, explainable, and cheaper defense approach called AttentionDefense. Our research suggests that the attention mechanism is an integral component in understanding and explaining how LMs respond to malicious input that is not captured in the semantic meaning of text embeddings. The proposed AttentionDefense is evaluated against existing jailbreak benchmark datasets. Ablation studies show that SLM-based AttentionDefense has equivalent or better jailbreak detection performance compared to text embedding-based classifiers and GPT-4 zero-shot detectors. To further validate the efficacy of the proposed approach, we generate a dataset of novel jailbreak variants of the existing benchmark dataset using a closed-loop LLM-based multi-agent system. We demonstrate that the proposed AttentionDefense approach performs robustly on this novel jailbreak dataset while existing approaches suffer in performance. Additionally, for practical purposes AttentionDefense is an ideal solution as it has the computation requirements of a small LM but the performance of a LLM detector.

Designing deep learning-based solutions is becoming a race for training deeper models with a greater number of layers. While a large-size deeper model could provide competitive accuracy, it creates a lot of logistical challenges and unreasonable resource requirements during development and deployment. This has been one of the key reasons for deep learning models not being excessively used in various production environments, especially in edge devices. There is an immediate requirement for optimizing and compressing these deep learning models, to enable on-device intelligence. In this research, we introduce a black-box framework, Deeplite Neutrino^{TM} for production-ready optimization of deep learning models. The framework provides an easy mechanism for the end-users to provide constraints such as a tolerable drop in accuracy or target size of the optimized models, to guide the whole optimization process. The framework is easy to include in an existing production pipeline and is available as a Python Package, supporting PyTorch and Tensorflow libraries. The optimization performance of the framework is shown across multiple benchmark datasets and popular deep learning models. Further, the framework is currently used in production and the results and testimonials from several clients are summarized.

Question answering is one of the primary challenges of natural language understanding. In realizing such a system, providing complex long answers to questions is a challenging task as opposed to factoid answering as the former needs context disambiguation. The different methods explored in the literature can be broadly classified into three categories namely: 1) classification based, 2) knowledge graph based and 3) retrieval based. Individually, none of them address the need of an enterprise wide assistance system for an IT support and maintenance domain. In this domain, the variance of answers is large ranging from factoid to structured operating procedures; the knowledge is present across heterogeneous data sources like application specific documentation, ticket management systems and any single technique for a general purpose assistance is unable to scale for such a landscape. To address this, we have built a cognitive platform with capabilities adopted for this domain. Further, we have built a general purpose question answering system leveraging the platform that can be instantiated for multiple products, technologies in the support domain. The system uses a novel hybrid answering model that orchestrates across a deep learning classifier, a knowledge graph based context disambiguation module and a sophisticated bag-of-words search system. This orchestration performs context switching for a provided question and also does a smooth hand-off of the question to a human expert if none of the automated techniques can provide a confident answer. This system has been deployed across 675 internal enterprise IT support and maintenance projects.

With an abundance of research papers in deep learning, adoption and reproducibility of existing works becomes a challenge. To make a DL developer life easy, we propose a novel system, DARVIZ, to visually design a DL model using a drag-and-drop framework in an platform agnostic manner. The code could be automatically generated in both Caffe and Keras. DARVIZ could import (i) any existing Caffe code, or (ii) a research paper containing a DL design; extract the design, and present it in visual editor.

With an abundance of research papers in deep learning, reproducibility or adoption of the existing works becomes a challenge. This is due to the lack of open source implementations provided by the authors. Even if the source code is available, then re-implementing research papers in a different library is a daunting task. To address these challenges, we propose a novel extensible approach, DLPaper2Code, to extract and understand deep learning design flow diagrams and tables available in a research paper and convert them to an abstract computational graph. The extracted computational graph is then converted into execution ready source code in both Keras and Caffe, in real-time. An arXiv-like website is created where the automatically generated designs is made publicly available for 5,000 research papers. The generated designs could be rated and edited using an intuitive drag-and-drop UI framework in a crowd sourced manner. To evaluate our approach, we create a simulated dataset with over 216,000 valid deep learning design flow diagrams using a manually defined grammar. Experiments on the simulated dataset show that the proposed framework provide more than 93% accuracy in flow diagram content extraction.

In Sanskrit, small words (morphemes) are combined to form compound words through a process known as Sandhi.Sandhi splitting is the process of splitting a given compound word into its constituent morphemes.Although rules governing word splitting exists in the language, it is highly challenging to identify the location of the splits in a compound word.Though existing Sandhi splitting systems incorporate these pre-defined splitting rules, they have a low accuracy as the same compound word might be broken down in multiple ways to provide syntactically correct splits.In this research, we propose a novel deep learning architecture called Double Decoder RNN (DD-RNN), which (i) predicts the location of the split(s) with 95% accuracy, and (ii) predicts the constituent words (learning the Sandhi splitting rules) with 79.5% accuracy, outperforming the state-of-art by 20%.Additionally, we show the generalization capability of our deep learning model, by showing competitive results in the problem of Chinese word segmentation, as well.

Latent fingerprint identification is of critical importance in criminal investigation. FBI's Next Generation Identification program demands latent fingerprint identification to be performed in lights-out mode, with very little or no human intervention. However, the performance of an automated latent fingerprint identification is limited due to imprecise automated feature (minutiae) extraction, specifically due to noisy ridge pattern and presence of background noise. In this paper, we propose a novel descriptor based minutiae detection algorithm for latent fingerprints. Minutia and non-minutia descriptors are learnt from a large number of tenprint fingerprint patches using stacked denoising sparse autoencoders. Latent fingerprint minutiae extraction is then posed as a binary classification problem to classify patches as minutia or non-minutia patch. Experiments performed on the NIST SD-27 database shows promising results on latent fingerprint matching.

Over the years, automatic gender recognition has been used in many applications. However, limited research has been done on analyzing gender recognition across ethnicity scenario. This research aims at studying the performance of discriminant functions including Principal Component Analysis, Linear Discriminant Analysis and Subclass Discriminant Analysis with the availability of limited training database and unseen ethnicity variations. The experiments are performed on a heterogeneous database of 8112 images that includes variations in illumination, expression, minor pose and ethnicity. Contrary to existing literature, the results show that PCA provides comparable but slightly better performance compared to PCA+LDA, PCA+SDA and PCA+SVM. The results also suggest that linear discriminant functions provide good generalization capability even with limited number of training samples, principal components and with cross-ethnicity variations.

This research presents a forensics application of matching two latent fingerprints. In crime scene settings, it is often required to match multiple latent fingerprints. Unlike matching latent with inked or live fingerprints, this research problem is very challenging and requires proper analysis and attention. The contribution of this paper is three fold: (i) a comparative analysis of existing algorithms is presented for this application, (ii) fusion and context switching frameworks are presented to improve the identification performance, and (Hi) a multi-latent fingerprint database is prepared. The experiments highlight the need for improved feature extraction and processing methods and exhibit large scope of improvement in this important research problem.