In-Context Learning (ICL) has become a powerful paradigm that enables LLMs to perform a wide range of tasks without task-specific fine-tuning. However, the effectiveness of ICL heavily depends on the quality of exemplar selection. In particular, for structured prediction tasks such as semantic parsing, existing ICL selection strategies often overlook structural alignment, leading to suboptimal performance and poor generalization. To address this issue, we propose a novel two-stage exemplar selection strategy that achieves a strong balance between efficiency, generalizability, and performance. First, we fine-tune a BERT-based retriever using structure-aware supervision, guiding it to select exemplars that are both semantically relevant and structurally aligned. Then, we enhance the retriever with a plug-in module, which amplifies syntactically meaningful information in the hidden representations. This plug-in is model-agnostic, requires minimal overhead, and can be seamlessly integrated into existing pipelines. Experiments on four benchmarks spanning three semantic parsing tasks demonstrate that our method consistently outperforms existing baselines with multiple recent LLMs as inference-time models.

Probabilistic programs provide an expressive representation language for generative models. Given a probabilistic program, we are interested in the task of posterior inference: estimating a latent variable given a set of observed variables. Existing techniques for inference in probabilistic programs often require choosing many hyper-parameters, are computationally expensive, and/or only work for restricted classes of programs. Here we formulate inference as masked language modeling: given a program, we generate a supervised dataset of variables and assignments, and randomly mask a subset of the assignments. We then train a neural network to unmask the random values, defining an approximate posterior distribution. By optimizing a single neural network across a range of programs we amortize the cost of training, yielding a "foundation" posterior able to do zero-shot inference for new programs. The foundation posterior can also be fine-tuned for a particular program and dataset by optimizing a variational inference objective. We show the efficacy of the approach, zero-shot and fine-tuned, on a benchmark of STAN programs.

We're Cruise, a self-driving service designed for the cities we love. We're building the world's most advanced, self-driving vehicles to safely connect people to the places, things, and experiences they care about. We believe self-driving vehicles will help save lives, reshape cities, give back time in transit, and restore freedom of movement for many. Cruisers have the opportunity to grow and develop while learning from leaders at the forefront of their fields. With a culture of internal mobility, there's an opportunity to thrive in a variety of disciplines.

Machine Learning India (MLI) is a thriving community of over 450,000 ardent - artificial intelligence enthusiasts across India and the globe. We at MLI, believe that India has the potential to position itself among leaders, on the global technology map. The goal of MLI is to reduce the skill-gap in India, by creating a vibrant AI ecosystem and talent pool; thereby leading our country to have a significant take in the global AI revolution. To pursue the same, we intend to democratize quality technical education, resources and opportunities and make it available to all.

Hands-on Machine Learning With Python: An online, 12-hour workshop focused on machine learning algorithms and their application. Targeted at: Engineering (all streams) as well as science students, who are looking forward to get started with machine learning. Instructor profile: A certified data scientist, with a demonstrated history of working in the information technology and services industry. Note: We will be limiting the participants to 50. About Us: Founded in 2018, Machine Learning India (MLI), is a thriving community of 350,000 passionate technologists across India and the globe.

Start Date: Tuesday 21st April 2020 The updated certificate now includes 25 lecture weeks, our new Partnership with NAG Numerical NAG (Numerical Algorithms Group), additional practical lab sessions, an extended module 1 on Supervised Learning, new topic updates on Cloud Computing, Natural Language Processing, Practicalities of Neural Networks: CNN, Advanced Practicalities of Neural Networks: Generative NN, and a new full module on Times Series. Quantitative finance is moving into a new era. Traditional quant skills are no longer adequate to deal with the latest challenges in finance. The Machine Learning Institute Certificate offers candidates the chance to upgrade their skill set by combining academic rigour with practical industry insight. The Machine Learning Institute Certificate in Finance (MLI) is a comprehensive six-month part-time course, with weekly live lectures in London or globally online.

Investors dropped 681 million into A.I.-centric startups in Silicon Valley last year. This year, the number will likely reach 1.2 billion. Five years ago, total A.I. investment spiked at roughly 150 million. This is how Silicon Valley works: When something new is hyped and seems to have investor trust, everybody jumps on the train without asking, "Where does this train go?" The truth is that artificial intelligence does not exist yet, and most companies claiming to have A.I. technology are arrogantly re-selling an old concept of machine learning -- a technology that was first introduced in 1959 but which truly started to take off in the 1990s.

Investors dropped 681 million into A.I.-centric startups in Silicon Valley last year. This year, the number will likely reach 1.2 billion. Five years ago, total A.I. investment spiked at roughly 150 million. This is how Silicon Valley works: When something new is hyped and seems to have investor trust, everybody jumps on the train without asking, "Where does this train go?" The truth is that artificial intelligence does not exist yet, and most companies claiming to have A.I. technology are arrogantly re-selling an old concept of machine learning -- a technology that was first introduced in 1959 but which truly started to take off in the 1990s.

Investors dropped 681 million into A.I.-centric startups in the Valley last year. This year the number would reach 1.2 billion. Five years ago total A.I. investment piked at roughly 150 million. This is how Silicon Valley works: when there is a new hype that seems to have investor trust everybody jumps on the train without asking any questions. No one asked: "Where does this train go?"

The recent success stories of machine learning (ML) driven applications have created an increasing demand for scalable ML solutions. Nonetheless, ML researchers often prefer to code their solutions in statistical computing languages such as MATLAB or R, as these languages allow them to code in fewer lines using syntax that resembles high-level pseudocode. MATLAB and R allow researchers to avoid low-level implementation details, leading to quickly developed prototypes that are often sufficient for small scale exploration. However, these prototypes are typically ad-hoc, non-robust, and non-scalable implementations. In contrast, industrial implementations of these solutions often require a relatively heavy amount of development effort and are difficult to change once implemented.