Geetanjali Rathee

Aparna Singh, Geetanjali Rathee, Chaker Abdelaziz Kerrache et al.

The very high growth of Intelligent Transportation Systems (ITS) has generated an urgent requirement for secure, effective, and context-aware data sharing mechanisms, especially over heterogeneous and geographically dispersed settings. This work suggests a new architecture that combines a relay chain-driven encryption system with a modified Ciphertext-Policy Attribute-Based Encryption (CP-ABE) scheme to tackle the double impediment of dynamic access and low-latency communication. The model proposes a context-aware smart contract on a worldwide relay chain that checks against data properties, including event type, time, and geographical region, to specify the suitable level of encryption policy. From such relay-directed judgment, On-Board Units (OBUs) encrypt data end-to-end by utilising CP-ABE and store ciphertext inside localised regional blockchains, preventing dependence on symmetric encryption or off-chain storage. High-sensitivity events are secured with firm, multi-attribute access rules, whereas common updates use light policies to help reduce processing burdens. The crypto system also adds traceability and low-latency revocation, with global enforcement managed through the relay chain. This distributed, scalable model provides a proper balance between responsiveness in real time and security and is extremely apt for next-gen vehicular networks that function across multi-jurisdictional domains.

Kartik Jangra, Aman Kumar Singh, Yashwani Mann et al.

Recent advancements in vision-language models have achieved remarkable results in making language models understand vision inputs. However, a unified approach to align these models across diverse tasks such as image captioning and visual question answering remains a challenge. Existing methods either require very big language models or very big datasets which is not efficient in utilizing existing models. This paper addresses this gap and devises a training strategy of auto-regressive vision-language models, to unify vision-language tasks like image-captioning and visual question answering. We propose four training stages for aligning the vision model with the language model, in other words, the language model is given an ability to process visual inputs. We also devise different attention masks for training transformer-based language models that improve the quality of visual features. Further, we introduce some findings, 1) the attention mask should not be applied on visual inputs, 2) the Language model converges faster on AI- generated data, 3) More work should be done in the alignment stage during the pre-training of the model, 4) the model can easily adapt to any downstream tasks like visual question answering on healthcare datasets like PathVQA. After training the model for one epoch for all the stages, it outperforms large models like VILA-13 billion models on common benchmarks like CIDEr scores on COCO and Flickr30k datasets and achieves very close scores to GIT-2 on the same dataset despite being a much smaller model trained on a much smaller dataset. All of the training is done using best practices available like multi- GPU parallel training, lower-precision training with 16-bit float numbers, faster attention (SDPA), and gradient accumulation, and completed the training within 12 hours.

Satvik Garg, Pradyumn Pundir, Geetanjali Rathee et al.

Model deployment in machine learning has emerged as an intriguing field of research in recent years. It is comparable to the procedure defined for conventional software development. Continuous Integration and Continuous Delivery (CI/CD) have been shown to smooth down software advancement and speed up businesses when used in conjunction with development and operations (DevOps). Using CI/CD pipelines in an application that includes Machine Learning Operations (MLOps) components, on the other hand, has difficult difficulties, and pioneers in the area solve them by using unique tools, which is typically provided by cloud providers. This research provides a more in-depth look at the machine learning lifecycle and the key distinctions between DevOps and MLOps. In the MLOps approach, we discuss tools and approaches for executing the CI/CD pipeline of machine learning frameworks. Following that, we take a deep look into push and pull-based deployments in Github Operations (GitOps). Open exploration issues are also identified and added, which may guide future study.