​Using audio and text embeddings jointly for Keyword Spotting (KWS) has shown high-quality results, but the key challenge of how to semantically align two embeddings for multi-word keywords of different sequence lengths remains largely unsolved. In this paper, we propose an audio-text-based end-to-end model architecture… Read More »Matching Latent Encoding for Audio-Text based Keyword Spotting Apple Machine Learning Research

​*=Equal Contributors We consider online learning problems in the realizable setting, where there is a zero-loss solution, and propose new Differentially Private (DP) algorithms that obtain near-optimal regret bounds. For the problem of online prediction from experts, we design new algorithms that obtain near-optimal regret… Read More »Near-Optimal Algorithms for Private Online Optimization in the Realizable Regime Apple Machine Learning Research

​This paper focuses on long-tailed object detection in the semi-supervised learning setting, which poses realistic challenges, but has rarely been studied in the literature. We propose a novel pseudo-labeling-based detector called CascadeMatch. Our detector features a cascade network architecture, which has multi-stage detection heads with… Read More »Semi-Supervised and Long-Tailed Object Detection with CascadeMatch Apple Machine Learning Research

​This paper presents an extension to train end-to-end Context-Aware Transformer Transducer ( CATT ) models by using a simple, yet efficient method of mining hard negative phrases from the latent space of the context encoder. During training, given a reference query, we mine a number… Read More »Approximate Nearest Neighbor Phrase Mining for Contextual Speech Recognition Apple Machine Learning Research

​The techniques for 3D indoor scene capturing are widely used, but the meshes produced leave much to be desired. In this paper, we propose “RoomDreamer”, which leverages powerful natural language to synthesize a new room with a different style. Unlike existing image synthesis methods, our… Read More »RoomDreamer: Text-Driven 3D Indoor Scene Synthesis with Coherent Geometry and Texture Apple Machine Learning Research

​ There is a lot of potentials for conversational generative AI to help medical professionals, but so far, the research has only focused on text. While advances in multi-modal conversational AI have been rapid because of billions of publicly available image-text pairings, such general-domain vision-language… Read More »Microsoft AI Unveils LLaVA-Med: An Efficiently Trained Large Language and Vision Assistant Revolutionizing Biomedical Inquiry, Delivering Advanced Multimodal Conversations in Under 15 Hours Dhanshree Shripad Shenwai Artificial Intelligence Category – MarkTechPost

​ Google has introduced the Secure AI Framework (SAIF), a conceptual framework that establishes clear industry security standards for building and deploying AI systems responsibly. SAIF draws inspiration from security best practices in software development and incorporates an understanding of security risks specific to AI… Read More »Google AI Introduces a New Secure AI Framework (SAIF): A Conceptual Framework for Ensuring the Security of AI Systems Niharika Singh Artificial Intelligence Category – MarkTechPost

​ Text-driven video editing aims to create new videos out of text prompts and existing video material without any manual labor. This technology has the potential to substantially impact various industries, including social media content, marketing, and advertising. The modified films must accurately reflect the… Read More »Meet ControlVideo: A Novel AI Method For Text-Driven Video Editing Aneesh Tickoo Artificial Intelligence Category – MarkTechPost

​ Generative Artificial Intelligence has become increasingly popular in the past few months. Being a subset of AI, it enables Large Language Models (LLMs) to generate new data by learning from massive amounts of available textual data. LLMs understand and follow user intentions and instructions… Read More »Meet Video-LLaMA: A Multi-Modal Framework that Empowers Large Language Models (LLMs) with the Capability of Understanding both Visual and Auditory Content in the Video Tanya Malhotra Artificial Intelligence Category – MarkTechPost

​ When it comes to natural language processing (NLP) tasks, large language models (LLM) trained on massive online datasets perform exceptionally well. Segment Anything Model (SAM) has shown outstanding zero-shot localization abilities in computer vision (CV) by scaling up data.  Unfortunately, SAM cannot produce semantic… Read More »A New AI Research Introduces Recognize Anything Model (RAM): A Robust Base Model For Image Tagging Tanushree Shenwai Artificial Intelligence Category – MarkTechPost