Posts in category: Microsoft AI

Members of the research community at Microsoft work continuously to advance their respective fields. Abstracts brings its audience to the cutting edge with them through short, compelling conversations about new and noteworthy achievements.

In this episode, Principal Researcher Alessandro Sordoni joins host Gretchen Huizinga to discuss “Joint Prompt Optimization of Stacked LLMs using Variational Inference.” In the paper, which was accepted at the 2023 Conference on Neural Information Processing Systems (NeurIPS), Sordoni and his coauthors introduce Deep Language Networks, or DLNs, an architecture that treats large language models as layers within a network and natural language prompts as each layer’s learnable parameters.

Opens in a new tab

The post Abstracts: December 11, 2023 appeared first on Microsoft Research.

Read More

Microsoft is proud to sponsor the 37th Conference on Neural Information Processing Systems (NeurIPS 2023). This interdisciplinary forum brings together experts in machine learning, neuroscience, statistics, optimization, computer vision, natural language processing, life sciences, natural sciences, social sciences, and other adjacent fields. We are pleased to share that Microsoft has over 100 accepted papers and is offering 18 workshops at NeurIPS 2023. 

This year’s conference includes three papers from Microsoft that were chosen for oral presentations, which feature groundbreaking concepts, methods, or applications, addressing pressing issues in the field. Additionally, our spotlights posters, also highlighted below, have been carefully curated by conference organizers, exhibiting novelty, technical rigor, and the potential to significantly impact the landscape of machine learning. This blog post celebrates those achievements.

Oral Presentations

Bridging Discrete and Backpropagation: Straight-Through and Beyond

Gradient computations are pivotal in deep learning’s success, yet they predominantly depend on backpropagation, a technique limited to continuous variables. The paper Bridging Discrete and Backpropagation: Straight-Through and Beyond, tackles this limitation. It introduces ReinMax, extending backpropagation’s capability to estimate gradients for models incorporating discrete variable sampling. Within extensive experiments of this study, ReinMax demonstrates consistent and significant performance gain over the state of the art. More than just a practical solution, the paper sheds light on existing deep learning practices. It elucidates that the ‘Straight-Through’ method, once considered merely a heuristic trick, is actually a viable first-order approximation for the general multinomial case. Correspondingly, ReinMax achieves second-order accuracy in this context without the complexities of second-order derivatives, thus having negligible computation overheads. 

MICROSOFT RESEARCH PODCAST

Intern Insights: Dr. Madeleine Daepp with Jennifer Scurrell and Alejandro Cuevas

In this episode, PhD students Jennifer Scurrell and Alejandro Cuevas talk to Senior Researcher Dr. Madeleine Daepp. They discuss the internship culture at Microsoft Research, from opportunities to connect with researchers to the teamwork they say helped make it possible for them to succeed, and the impact they hope to have with their work.

Listen now

Opens in a new tab

The MineRL BASALT Competition on Learning from Human Feedback

The growth of deep learning research, including its incorporation into commercial products, has created a new challenge: How can we build AI systems that solve tasks when a crisp, well-defined specification is lacking? To encourage research on this important class of techniques, researchers from Microsoft led The MineRL BASALT Competition on Learning from Human Feedback (opens in new tab), an update to a contest first launched in 2021 (opens in new tab) by researchers at the University of California-Berkeley and elsewhere. The challenge of this competition was to complete fuzzy tasks from English language descriptions alone, with emphasis on encouraging different ways of learning from human feedback as an alternative to a traditional reward signal. 

The researchers designed a suite of four tasks in Minecraft for which writing hardcoded reward functions would be difficult. These tasks are defined by natural language: for example, “create a waterfall and take a scenic picture of it”, with additional clarifying details. Participants must train a separate agent for each task. Agents are then evaluated by humans who have read the task description.

The competition aimed to encourage development of AI systems that do what their designers intended, even when the intent cannot be easily formalized. Besides allowing AI to solve more tasks, this can also enable more effective regulation of AI systems, as well as making progress on value alignment problems, in which the specified objectives of an AI agent differ from those of its users.

Related

DecodingTrust: A Comprehensive Assessment of Trustworthiness in GPT Models

This comprehensive evaluation platform aims to answer the question: How trustworthy are generative pre-trained transformer (GPT) models? In DecodingTrust: A Comprehensive Assessment of Trustworthiness in GPT Models, researchers focus specifically on GPT-4, GPT-3.5, and a series of open LLMs. They consider diverse perspectives, including toxicity, stereotype bias, adversarial robustness, out-of-distribution robustness, robustness on adversarial demonstrations, privacy, machine ethics, and fairness.

The researchers’ evaluations identified previously unpublished vulnerabilities relating to trustworthiness. The team worked with Microsoft product groups to confirm that the potential vulnerabilities identified do not impact current customer-facing services. This is in part true because finished AI applications apply a range of mitigation approaches to address potential harms that may occur at the model level of the technology. They also shared their findings with GPT’s developer, OpenAI, which has noted the potential vulnerabilities in the system cards for relevant models.

This research aims to encourage others in the research community to utilize and build upon this work, potentially pre-empting adversaries who would exploit vulnerabilities to cause harm. To facilitate collaboration, the benchmark code is very extensible and easy to use: a single command is sufficient to run the complete evaluation on a new model.

Spotlight Posters

Differentially Private Approximate Near Neighbor Counting in High Dimensions

Differential privacy (DP) is a widely used tool for preserving the privacy of sensitive personal information. It allows a data structure to provide approximate answers to queries about the data it holds, while ensuring that the removal or addition of a single database entry does not significantly affect the outcome of any analysis.

Range counting (counting the number of data points falling into a given query ball) under differential privacy has been studied extensively. However, current algorithms for this problem come with challenges. One class of algorithms suffers from an additive error that is a fixed polynomial in the number of points. Another class of algorithms allows for polylogarithmic additive error, but the error grows exponentially in the dimension. To achieve the latter, the problem is relaxed to allow a “fuzzy” definition of the range boundary, e.g., a count of the points in a ball of radius r might also include points in a ball of radius cr for some c > 1.

In Differentially Private Approximate Near Neighbor Counting in High Dimensions, researchers present an efficient algorithm that offers a sweet spot between these two classes. The algorithm has an additive error that is an arbitrary small power of the data set size, depending on how fuzzy the range boundary is, as well as a small (1 + o(1)) multiplicative error. Crucially, the amount of noise added has no dependence on the dimension. This new algorithm introduces a variant of Locality-Sensitive Hashing, utilizing it in a novel manner.

Exposing Attention Glitches with Flip-Flop Language Modeling

Why do large language models sometimes output factual inaccuracies and exhibit erroneous reasoning? The brittleness of these models, particularly when executing long chains of reasoning, seems to be an inevitable price to pay for their advanced capabilities of coherently synthesizing knowledge, pragmatics, and abstract thought.

To help make sense of this fundamentally unsolved problem, Exposing Attention Glitches with Flip-Flop Language Modeling identifies and analyzes the phenomenon of attention glitches, in which the Transformer architecture’s inductive biases intermittently fail to capture robust reasoning. To isolate the issue, the researchers introduce flip-flop language modeling (FFLM), a parametric family of synthetic benchmarks designed to probe the extrapolative behavior of neural language models. This simple generative task requires a model to copy binary symbols over long-range dependencies, ignoring the tokens in between. This research shows how Transformer FFLMs suffer from a long tail of sporadic reasoning errors, some of which can be eliminated using various regularization techniques. The preliminary mechanistic analyses show why the remaining errors may be very difficult to diagnose and resolve. The researchers hypothesize that attention glitches account for some of the closed-domain errors occurring in natural LLMs.

In-Context Learning Unlocked for Diffusion Models

An emergent behavior of large language models (LLMs) is the ability to learn from context, or in-context learning. With a properly designed prompt structure and in-context learning, LLMs can combine the pre-training of multiple language tasks and generalize well to previously unseen tasks. While in-context learning has been extensively studied in natural language processing (NLP), its applications in the field of computer vision are still limited.

In-Context Learning Unlocked for Diffusion Models presents Prompt Diffusion, a framework for enabling in-context learning in diffusion-based generative models. Given a pair of task-specific example images and text guidance, this model understands the underlying task and performs the same task on a new query image following the text guidance. To achieve this, the researchers propose a vision-language prompt that can model a wide range of vision-language tasks, and a diffusion model that takes it as input. The diffusion model is trained jointly over six different tasks using these prompts. The resulting Prompt Diffusion model is the first diffusion-based vision-language foundation model capable of in-context learning. It demonstrates high-quality in-context generation on the trained tasks and generalizes to new, unseen vision tasks with their respective prompts. This model also shows compelling text-guided image editing results.

Optimizing Prompts for Text-to-Image Generation

Generative foundation models can be prompted to follow user instructions, including language models and text-to-image models. Well-designed prompts can guide text-to-image models to generate amazing images. However, the performant prompts are often model-specific and misaligned with user input. Instead of laborious human engineering, Optimizing Prompts for Text-to-Image Generation proposes prompt adaptation, a general framework that automatically adapts original user input to model-preferred prompts.

The researchers use reinforcement learning to explore better prompts with a language model. They define a reward function that encourages the policy network (i.e., language model) to generate more aesthetically pleasing images while preserving the original user intentions. Experimental results on Stable Diffusion show that this method outperforms manual prompt engineering in terms of both automatic metrics and human preference ratings. Reinforcement learning further boosts performance, especially on out-of-domain prompts.

Pareto Frontiers in Neural Feature Learning: Data, Compute, Width, and Luck

Algorithm design in deep learning can appear to be more like “hacking” than an engineering practice. There are numerous architectural choices and training heuristics, which can often modulate model performance and resource costs in unpredictable and entangled ways. As a result, when training large-scale neural networks (such as state-of-the-art language models), algorithmic decisions and resource allocations are foremost empirically-driven, involving the measurement and extrapolation of scaling laws. A precise mathematical understanding of this process is elusive, and cannot be explained by statistics or optimization in isolation.

In Pareto Frontiers in Neural Feature Learning: Data, Compute, Width, and Luck, researchers from Microsoft, Harvard, and the University of Pennsylvania explore these algorithmic intricacies and tradeoffs through the lens of a single synthetic task: the finite-sample sparse parity learning problem. In this setting, the above complications are not only evident, but also provable: intuitively, due to the task’s computational hardness, a neural network needs a sufficient combination of resources (“data × model size × training time × luck”) to succeed. This research shows that standard algorithmic choices in deep learning give rise to a Pareto frontier, in which successful learning is “bought” with interchangeable combinations of these resources. They show that algorithmic improvements on this toy problem can transfer to the real world, improving the data-efficiency of neural networks on small tabular datasets.

PDE-Refiner: Achieving Accurate Long Rollouts with Neural PDE Solvers

Time-dependent partial differential equations (PDEs) are ubiquitous in science and engineering. The high computational cost of traditional solution techniques has spurred increasing interest in deep neural network based PDE surrogates. The practical utility of such neural PDE solvers depends on their ability to provide accurate, stable predictions over long time horizons, which is a notoriously hard problem.

PDE-Refiner: Achieving Accurate Long Rollouts with Neural PDE Solvers presents a large-scale analysis of common temporal rollout strategies, identifying the neglect of non-dominant spatial frequency information, often associated with high frequencies in PDE solutions, as the primary pitfall limiting stable, accurate rollout performance. Motivated by recent advances in diffusion models, the researchers developed PDE-Refiner, a novel model class that enables more accurate modeling of all frequency components via a multistep refinement process. They validate PDE-Refiner on challenging benchmarks of complex fluid dynamics, demonstrating stable and accurate rollouts that consistently outperform state-of-the-art models, including neural, numerical, and hybrid neural-numerical architectures. They also demonstrate that PDE-Refiner greatly enhances data efficiency, since the denoising objective implicitly induces a novel form of spectral data augmentation. Finally, PDE-Refiner’s connection to diffusion models enables an accurate and efficient assessment of the model’s predictive uncertainty, allowing researchers to estimate when the surrogate becomes inaccurate.

Should I Stop or Should I Go: Early Stopping with Heterogeneous Populations

Randomized experiments are the gold-standard method of determining causal effects, whether in clinical trials to evaluate medical treatments or in A/B tests to evaluate online product offerings. But randomized experiments often need to be stopped prematurely when the treatment or test causes an unintended harmful effect. Existing methods that determine when to stop an experiment early are typically applied to the data in aggregate and do not account for treatment effect heterogeneity.

Should I Stop or Should I Go: Early Stopping with Heterogeneous Populations examines the early stopping of experiments for harm on heterogeneous populations. The paper shows that current methods often fail to stop experiments when the treatment harms a minority group of participants. The researchers use causal machine learning to develop Causal Latent Analysis for Stopping Heterogeneously (CLASH), the first broadly-applicable method for heterogeneous early stopping. They demonstrate CLASH’s performance on simulated and real data and show that it yields effective early stopping for both clinical trials and A/B tests.

Survival Instinct in Offline Reinforcement Learning

In offline reinforcement learning (RL), an agent optimizes its performance given an offline dataset. Survival Instinct in Offline Reinforcement Learning presents a novel observation: on many benchmark datasets, offline RL can produce well-performing and safe policies even when trained with “wrong” reward labels, such as those that are zero everywhere or are negatives of the true rewards. This phenomenon cannot be easily explained by offline RL’s return maximization objective. Moreover, it gives offline RL a degree of robustness that is uncharacteristic of its online RL counterparts, which are known to be sensitive to reward design.

This research demonstrates that this surprising robustness property is attributable to an interplay between the notion of pessimism in offline RL algorithms and a certain bias implicit in common data collection practices. This work shows that this pessimism endows the agent with a “survival instinct”, i.e., an incentive to stay within the data support in the long term, while the limited and biased data coverage further constrains the set of survival policies. The researchers argue that the survival instinct should be taken into account when interpreting results from existing offline RL benchmarks and when creating future ones.

Timewarp: Transferable Acceleration of Molecular Dynamics by Learning Time-Coarsened Dynamics

Molecular dynamics (MD) is a well-established technique for simulating physical systems at the atomic level. When performed accurately, it provides unrivalled insight into the detailed mechanics of molecular motion, without the need for wet lab experiments. MD is often used to compute equilibrium properties, which requires sampling from an equilibrium distribution such as the Boltzmann distribution (opens in new tab). However, many important processes, such as binding and folding, occur over timescales of milliseconds or beyond, and cannot be efficiently sampled with conventional MD. Furthermore, new MD simulations need to be performed from scratch for each molecular system studied.

Timewarp: Transferable Acceleration of Molecular Dynamics by Learning Time-Coarsened Dynamics presents an enhanced sampling method which uses a normalizing flow as a proposal distribution in a Markov chain Monte Carlo method targeting the Boltzmann distribution. The flow is trained offline on MD trajectories and learns to make large steps in time, simulating the molecular dynamics of 10^5−10^6fs. Crucially, Timewarp is transferable between molecular systems: the researchers show that, once trained, Timewarp generalizes to unseen small peptides (2-4 amino acids), exploring their metastable states and providing wall-clock acceleration when sampling compared to standard MD. This new method constitutes an important step towards developing general, transferable algorithms for accelerating MD.

Opens in a new tab

The post NeurIPS 2023 highlights breadth of Microsoft’s machine learning innovation appeared first on Microsoft Research.

Read More

Generative AI has revolutionized how we create text and images. How about designing novel materials? We at Microsoft Research AI4Science are thrilled to announce MatterGen, our generative model that enables broad property-guided materials design.

The central challenge in materials science is to discover materials with desired properties, e.g., high Li-ion conductivity for battery materials. Traditionally, this has been done by first finding novel materials and then filtering down based on the application. This is like trying to create the image of a cat by first generating a million different images and then searching for the one with a cat. In MatterGen, we directly generate novel materials with desired properties, similar to how DALL·E 3 tackles image generation.  

MatterGen is a diffusion model specifically designed for generating novel, stable materials. MatterGen also has adapter modules that can be fine-tuned to generate materials given a broad range of constraints, including chemistry, symmetry, and properties. MatterGen generates 2.9 times more stable (≤ 0.1 eV/atom of our training + test data convex hull), novel, unique structures than a SOTA model (CDVAE). It also generates structures 17.5 times closer to energy local minimum. MatterGen can directly generate materials satisfying desired magnetic, electronic, mechanical properties via classifier-free guidance. We verify generated materials with DFT-based workflows. 

Additionally, MatterGen can keep generating novel materials that satisfy a target property like high bulk modulus while screening methods instead saturate due to the exhaustion of materials in the database.

MatterGen can also generate materials given target chemical systems. It outperforms substitution and random structure search baselines equipped with MLFF filtering, especially in challenging 5-element systems. MatterGen also generates structures given target space groups. Finally, we tackle the multi-property materials design problem of finding low-supply-chain risk magnets. MatterGen proposes structures that have both high magnetic density and a low supply-chain risk chemical composition. 

We believe MatterGen is an important step forward in AI for materials design. Our results are currently verified via DFT, which has many known limitations. Experimental verification remains the ultimate test for real-word impact, and we hope to follow up with more results. 

None of this would be possible without the highly collaborative work between Andrew Fowler, Claudio Zeni, Daniel Zügner, Matthew Horton, Robert Pinsler, Ryota Tomioka, Tian Xie and our amazing interns Xiang Fu, Sasha Shysheya, and Jonathan Crabbé, as well as Jake Smith, Lixin Sun and the entire AI4Science Materials Design team.  

We are also grateful for all the help from Microsoft Research, AI4Science, and Azure Quantum.

Opens in a new tab

The post MatterGen: Property-guided materials design appeared first on Microsoft Research.

Read More

This research paper was presented at the 2023 Conference on Empirical Methods in Natural Language Processing (opens in new tab) (EMNLP 2023), the premier conference on natural language processing and artificial intelligence.

As large language models (LLMs) models advance and their potential becomes increasingly apparent, an understanding is emerging that the quality of their output is directly related to the nature of the prompt that is given to them. This has resulted in the rise of prompting technologies, such as chain-of-thought (CoT) and in-context-learning (ICL), which facilitate an increase in prompt length. In some instances, prompts now extend to tens of thousands of tokens, or units of text, and beyond. While longer prompts hold considerable potential, they also introduce a host of issues, such as the need to exceed the chat window’s maximum limit, a reduced capacity for retaining contextual information, and an increase in API costs, both in monetary terms and computational resources.

To address these challenges, we introduce a prompt-compression method in our paper, “LLMLingua: Compressing Prompts for Accelerated Inference of Large Language Models (opens in new tab),” presented at EMNLP 2023 (opens in new tab). Using a well-trained small language model, such as GPT2-small or LLaMA-7B, LLMLingua identifies and removes unimportant tokens from prompts. This compression technique enables closed LLMs to make inferences from the compressed prompt. Although the token-level compressed prompts may be difficult for humans to understand, they prove highly effective for LLMs. This is illustrated in Figure 1.

LLMLingua’s method and evaluation

To develop LLMLingua’s framework, we employed a budget controller to balance the sensitivities of different modules in the prompt, preserving the language’s integrity. Our two-stage process involved course-grained prompt compression. We first streamlined the prompt by eliminating certain sentences and then individually compressed the remaining tokens. To preserve coherence, we employed an iterative token-level compression approach, refining the individual relationships between tokens. Additionally, we fine-tuned the smaller model to capture the distribution information from different closed LLMs by aligning it with the patterns in the LLMs’ generated data. We did this through instruction tuning.

To assess LLMLingua’s performance, we tested compressed prompts on four different datasets, GSM8K, BBH, ShareGPT, and Arxiv-March23, encompassing ICL, reasoning, summarization, and conversation. Our approach achieved impressive results, achieving up to 20x compression while preserving the original prompt’s capabilities, particularly in ICL and reasoning. LLMLingua also significantly reduced system latency.

During our test, we used LLaMA-7B as the small language model and GPT-3.5-Turbo-0301, one of OpenAI’s LLMs, as the closed LLM. The results show that LLMLingua maintains the original reasoning, summarization, and dialogue capabilities of the prompt, even at a maximum compression ratio of 20x, as reflected in the evaluation metric (EM) columns in Tables 1 and 2. At the same time, other compression methods failed to retain key semantic information in prompts, especially in logical reasoning details. For a more in-depth discussion of these results, refer to section 5.2 of the paper.

LLMLingua is robust, cost-effective, efficient, and recoverable

LLMLingua also showed impressive results across various small language models and different closed LLMs. When using GPT-2-small, LLMLingua achieved a strong performance score of 76.27 under the ¼-shot constraint, close to the LLaMA-7B’s result of 77.33 and surpassing the standard prompt results of 74.9. Similarly, even without aligning Claude-v1.3, one of the post powerful LLMs, LLMLingua’s score was 82.61 under the ½-shot constraint, outperforming the standard prompt result of 81.8.

LLMLingua also proved effective in reducing response length, leading to significant reductions in latency in the LLM’s generation process, with reductions ranging between 20 to 30 percent, as shown in Figure 2.

What makes LLMLingua even more impressive is its recoverability feature. When we used GPT-4 to restore the compressed prompts, it successfully recovered all key reasoning information from the full nine-step chain-of-thought (CoT) prompting, which enables LLMs to address problems through sequential intermediate steps. The recovered prompt was almost identical to the original, and its meaning was retained. This is shown in Tables 3 and 4.

Enhancing the user experience and looking ahead

LLMLingua is already proving its value through practical application. It has been integrated into LlamaIndex (opens in new tab), a widely adopted retrieval-augmented generation (RAG) framework. Currently, we are collaborating with product teams to reduce the number of tokens required in LLM calls, particularly for tasks like multi-document question-answering. Here, our goal is to significantly improve the user experience with LLMs. 

For the long-term, we have proposed LongLLMLingua, a prompt-compression technique designed for long-context scenarios, such as retrieval-augmented question-answering tasks in applications like chatbots, useful when information evolves dynamically over time. It’s also geared for tasks like summarizing online meetings. LongLLMLingua’s primary objective is to enhance LLMs’ ability to perceive key information, making it suitable for numerous real-world applications, notably information-based chatbots. We’re hopeful that this innovation paves the way for more sophisticated and user-friendly interactions with LLMs.

Learn more about our work on the LLMLingua (opens in new tab) page.

Opens in a new tab

The post LLMLingua: Innovating LLM efficiency with prompt compression appeared first on Microsoft Research.

Read More

Members of the research community at Microsoft work continuously to advance their respective fields. Abstracts brings its audience to the cutting edge with them through short, compelling conversations about new and noteworthy achievements.

In this episode, Xing Xie, a Senior Principal Research Manager at Microsoft Research, joins host Gretchen Huizinga to discuss “Evaluating General-Purpose AI with Psychometrics.” As AI capabilities move from task specific to more general purpose, the paper explores psychometrics, a subfield of psychology, as an alternative to traditional methods for evaluating model performance and for supporting consistent and reliable systems.

Opens in a new tab

The post Abstracts: December 6, 2023 appeared first on Microsoft Research.

Read More

These research papers were presented at the ACM Joint European Software Engineering Conference and Symposium on the Foundations of Software Engineering (opens in new tab) (ESEC/FSE 2023), a premier conference in the field of software engineering.

The practice of software development inevitably involves the challenge of handling bugs and various coding irregularities. These issues can become pronounced when developers engage in the common practice of copying and pasting code snippets from the web or other peer projects. While this approach might offer a quick solution, it can introduce a host of potential complications, including compilation issues, bugs, and even security vulnerabilities into the developer’s codebase.

To address this, researchers at Microsoft have been working to advance different aspects of the software development lifecycle, from code adaptation to automated bug detection and repair. At ESEC/FSE 2023 (opens in new tab), we introduced two techniques aimed at enhancing coding efficiency. AdaptivePaste utilizes a learning-based approach to adapt and refine pasted code snippets in an integrated development environment (IDE). InferFix is an end-to-end program repair framework designed to automate bug detection and resolution. This blog outlines these technologies.

Microsoft Research Podcast

AI Frontiers: Models and Systems with Ece Kamar

Ece Kamar explores short-term mitigation techniques to make these models viable components of the AI systems that give them purpose and shares the long-term research questions that will help maximize their value. 

Listen now

Opens in a new tab

AdaptivePaste: Intelligent copy-paste in IDE

A widespread practice among developers involves adapting pasted code snippets to specific use cases. However, current code analysis and completion techniques, such as masked language modeling and CodeT5, do not achieve an acceptable level of accuracy in identifying and adapting variable identifiers within these snippets to align them with the surrounding code. In the paper, “AdaptivePaste: Intelligent Copy-Paste in IDE,” we propose a learning-based approach to source code adaptation, aiming to capture meaningful representations of variable usage patterns. First, we introduce a specialized dataflow-aware de-obfuscation pretraining objective for pasted code snippet adaptation. Next, we introduce a transformer-based model of two variants: a traditional unidecoder and parallel-decoder model with tied weights.

The unidecoder follows a standard autoregressive decoder formulation, mapping each variable in the pasted snippet to a unique symbol in the context or declaring a new variable. The parallel decoder duplicates the decoder for each anonymized symbol in the anonymized pasted snippet, predicting names independently and factorizing the output distribution per symbol. This enables selective code snippet adaptation by surfacing model predictions above a specified threshold and outputting “holes” where uncertainty exists.

To establish a dataflow-aware de-obfuscation pretraining objective for pasted code snippet adaptation, we assigned mask symbols to variable identifiers at the granularity of whole code tokens. The pre-existing code context was unanonymized, allowing the model to attend to existing identifier names defined in scope.

Our evaluation of AdaptivePaste showed promising results. It successfully adapted Python source code snippets with 67.8 percent exact match accuracy. When we analyzed the impact of confidence thresholds on model predictions, we observed that the parallel decoder transformer model improves precision to 85.9 percent in a selective code adaptation setting.

InferFix: End-to-end program repair with LLMs

Addressing software defects accounts for a significant portion of development costs. To tackle this, the paper, “InferFix: End-to-End Program Repair with LLMs over Retrieval-Augmented Prompts,” introduces a program repair framework that combines the capabilities of a state-of-the-art static analyzer called Infer, a semantic retriever model called Retriever, and a transformer-based model called Generator to address crucial security and performance bugs in Java and C#.

The Infer static analyzer is used to reliably detect, classify, and locate critical bugs within complex systems through formal verification. The Retriever uses a transformer encoder model to search for semantically equivalent bugs and corresponding fixes in large datasets of known bugs. It’s trained using a contrastive learning objective to excel at finding relevant examples of the same bug type.

The Generator employs a 12 billion-parameter codex model, fine-tuned on supervised bug-fix data. To enhance its performance, the prompts provided to the Generator are augmented with bug type annotations, bug contextual information, and semantically similar fixes retrieved from an external nonparametric memory by the Retriever. The Generator generates the candidate to fix the bug.

To test InferFix, we curated a dataset called InferredBugs (opens in new tab), which is rich in metadata and comprises bugs identified through executing the Infer static analyzer on thousands of Java and C# repositories. The results are noteworthy. InferFix outperforms strong LLM baselines, achieving a top-1 accuracy of 65.6 percent in C# and an impressive 76.8 percent in Java on the InferredBugs dataset.

Looking ahead

With AdaptivePaste and InferFix, we hope to significantly streamline the coding process, minimizing errors and enhancing efficiency. This includes reducing the introduction of bugs when code snippets are added and providing automated bug detection, classification, and patch validation. We believe that these tools hold promise for an enhanced software development workflow, leading to reduced costs and an overall boost in project efficiency.

Looking ahead, the rapid advancement of LLMs like GPT-3.5 and GPT-4 has sparked our interest in exploring ways to harness their potential in bug management through prompt engineering and other methods. Our goal is to empower developers by streamlining the bug detection and repair process, facilitating a more robust and efficient development environment.

Opens in a new tab

The post Microsoft at ESEC/FSE 2023: AI techniques for a streamlined coding workflow appeared first on Microsoft Research.

Read More

NEW RESEARCH

Leveraging Large Language Models for Automated Proof Synthesis in Rust

Formal verification can probably guarantee the correctness of critical system software, but the high proof burden has long hindered its wide adoption. Recently, large language models (LLMs) have shown success in code analysis and synthesis. In a recent paper: Leveraging Large Language Models for Automated Proof Synthesis in Rust, researchers from Microsoft present a combination of LLMs and static analysis to synthesize invariants, assertions, and other proof structures for a Rust-based formal verification framework called Verus.

In a few-shot setting, GPT-4 demonstrates impressive logical ability in generating postconditions and loop invariants, especially when analyzing short code snippets. However, GPT-4 does not consistently retain and propagate the full context information needed by Verus, a task that can be straightforwardly accomplished through static analysis. Based on these observations, the researchers developed a prototype based on OpenAI’s GPT-4 model. This prototype decomposes the verification task into multiple smaller ones, iteratively queries GPT-4, and combines its output with lightweight static analysis. Evaluating the prototype with a developer in the automation loop on 20 vector-manipulating programs showed that it significantly reduces human effort in writing entry-level proof code.

MICROSOFT RESEARCH PODCAST

Intern Insights: Dr. Madeleine Daepp with Jennifer Scurrell and Alejandro Cuevas

In this episode, PhD students Jennifer Scurrell and Alejandro Cuevas talk to Senior Researcher Dr. Madeleine Daepp. They discuss the internship culture at Microsoft Research, from opportunities to connect with researchers to the teamwork they say helped make it possible for them to succeed, and the impact they hope to have with their work.

Listen now

Opens in a new tab

NEW RESEARCH

Don’t Forget the User: It’s Time to Rethink Network Measurements

The goal of network measurement is to characterize how and how well a network is performing. This has traditionally meant a focus on the bits and bytes — low-level network metrics such as latency and throughput, which have the advantage of being objective but are limited in representativeness and reach. In a recent paper: Don’t Forget the User: It’s Time to Rethink Network Measurements, researchers from Microsoft argue that users also provide a rich and largely untapped source of implicit and explicit signals that could complement and expand the coverage of traditional measurement methods. Implicit feedback leverages user actions to indirectly infer network performance and the resulting quality of user experience. Explicit feedback leverages user input, typically provided offline, to expand the reach of network measurement, especially for newer ones.

The researchers analyze example scenarios, including capturing implicit feedback through user actions such as the user (un)muting the mic or turning on/off the camera in a large-scale conferencing service. These techniques complement existing measurement methods and open a broad set of research directions, ranging from rethinking measurements tools, to designing user-centric networked systems and applications.

TECHNOLOGY UPDATE

Ghana 3D international telemedicine proof of concept study

A real-time 3D telemedicine system – leveraging Holoportation communication technology – was used to facilitate consultations with complex reconstructive patients prior, during, and after an overseas surgical collaboration. The system was used in a proof-of-concept clinic in November 2022 between Canniesburn Plastic Surgery Unit, UK, and the National Reconstructive Plastic Surgery and Burns Centre, Korle Bu Teaching Hospital, Ghana.

Four patients in Ghana were followed through their patient journey (mandibular ameloblastoma, sarcoma thigh, maxillary tumor, sarcoma back). A new report: Ghana 3D Telemedicine International MDT: A Proof-of-concept study details the responses of 13 participants (4 patients, 4 Ghana clinicians, 5 UK clinicians) completed feedback on the 3D multidisciplinary team (MDT). Outcome measures were rated highly with satisfaction 84.31/100, perceived benefit 4.54/5, overall quality 127.3/ 147, and usability 83.2/100. This data shows close alignment with that previously published on high income countries.

This novel technology has potential to enhance overseas surgical visits in low-to-middle income countries through improved planning, informed discussion with patients, expert consensus on complex cases, and fostering engagement with professionals who may be thousands of miles away.

The post Research Focus: Week of December 4, 2023 appeared first on Microsoft Research.

Read More

This research paper was presented at the Fourth African Human Computer Interaction Conference (opens in new tab) (AfriCHI 2023), the pan-African conference on interactive digital technology design.

Online health communities can be a lifeline for people seeking healthcare support, enabling them to share experiences, ask questions, and receive help. These are particularly vital in low-and-middle-income countries (LMICs), where access to quality healthcare can be limited and online health communities function as a doorway for receiving expert advice and accessing trustworthy content. One platform that is widely used for this purpose is WhatsApp due to its popularity and ability to host facilitated communities for specific groups, like patients affiliated with a particular clinic.

For all their benefits, online health communities also face challenges due to the myriad responsibilities and equal lack of support for facilitators, who must answer questions, respond to ongoing discussions, and review reports. Facilitation requires staying abreast of ongoing chat threads, verifying facts, and generally just being available. Given that most healthcare professionals already have a full day of in-person healthcare work, facilitation occurs during lunch breaks, evenings, and even mornings before the workday begins.

Our paper, “Can Large Language Models Support Medical Facilitation Work? A Speculative Analysis (opens in new tab),” presented at AfriCHI 2023 (opens in new tab), discusses research conducted in collaboration with the University of Washington, where we examined facilitated WhatsApp groups created for young people living with HIV in informal settlements in Kenya. Facilitation involved moderating chats, providing emotional support, conducting administrative tasks, sharing information, and resolving conflicts. Because many discussions occurred at night, facilitators struggled to keep up with the chats, often missing important questions or responding to them a few days after they were posted. Facilitators also found it difficult to defuse tensions, which occurred from time to time.

LLMs’ potential in supporting online health facilitators

To help resolve these challenges, we explored ways large language models (LLMs) could potentially support facilitators, for example, by flagging important messages and helping with content authoring. LLMs’ language translation capabilities and capacity to answer questions and summarize information made them great candidates for online heath communities, understanding that facilitators should always verify the content that LLMs create. To explore their potential, we tested their application on chat log data. We concluded that an LLM-enabled copilot could help facilitators in several ways, such as:

• Coproducing compelling content: LLMs could help facilitators create educational and informative content for group members. They can summarize frequently asked questions, patient stories, and best practices for managing chronic conditions.
• Summarizing messages: LLMs could summarize long discussions in the chat, making it easier for facilitators to get up to date and identify important issues. Summarization can also help participants who need to be offline and might otherwise miss important information.
• Providing recommendations: LLMs could help facilitators conduct research when answering questions. However, facilitators must exercise due diligence and verify any suggestions the LLM makes.
• Performing sentiment analysis: LLMs could flag potential trouble spots in messages, such as declines in mental health, tension among participants, harmful advice, and misinformation.
• Assigning badges: LLMs could assign badges to group members in recognition for participating in discussions, completing tasks, or achieving milestones. This could help to motivate and engage members.

Importance of human facilitation

While LLMs offer numerous potential benefits for healthcare facilitation, it’s important to consider their challenges and limitations. We strongly believe that LLMs should be used to augment, not replace, human facilitation. One crucial reason is that this technology cannot provide the emotional support essential in these groups. Another challenge involves the potential for bias and harm. LLMs are trained on massive datasets of text and code, which might contain harmful biases and stereotypes. Additionally, LLMs can produce errors when dealing with content from outside the training data, such as cultural backgrounds that are underrepresented in this data.

Our research shows that the benefits these groups provide lie beyond merely providing information. Their success, gauged by participation levels, perceived value by members, and adherence to medical protocols, is attributed not only to the facilitators’ expertise but also to their empathy, humor, and care. These are human qualities that LLMs cannot replace.

Looking forward

When used to augment and support existing medical professionals, LLMs show promise in healthcare solutions, such as those for patients with chronic diseases in LMICs. We recommend that future research and practice in this area prioritize the following:

• Developing and testing LLM-enabled copilot systems that are tailored to specific patient populations and online health communities.
• Ensuring that design supports medical professionals, taking special care to preserve their agency.
• Designing copilot systems so that users can easily evaluate output as well as identify and correct erroneous content.
• Developing guidelines and regulations to ensure quality and safety when using LLMs for healthcare purposes.

Overall, the use of LLMs to support the work of online health community facilitation is an exciting new area of research. By making the facilitators’ tasks easier, they can pave the way for groups supporting more patients, improve adherence to medical protocols, and enhance well-being. While our research focused on a specific type of WhatsApp group, the potential of LLMs reaches far beyond. These models have the potential to support facilitators of online health communities across a diverse range of platforms.

Opens in a new tab

The post Exploring LLMs’ potential to help facilitators enhance online healthcare communities appeared first on Microsoft Research.

Read More

This research is being presented at the Agami Summit 2023 (opens in new tab), an annual forum in Maharashtra, India, for innovation in the field of law and justice.

In one scenario of the future, such as the one imagined by Matt Welsh for the Association of Computing Machinery (ACM) (opens in new tab), AI will take the lead in coding while humans oversee the process. This shift will require people to take a supervisory role, focusing on high-level tasks while leaving the code details to AI. As we envision this transformation, we face a critical question: How can we reimagine software development to not just improve developer productivity but also to ensure software safety, reliability, and maintainability while keeping it personalized to developer preferences?

Realizing this outcome relies on AI and developers establishing a common understanding. While natural language can facilitate AI-developer interaction, it also introduces the potential for misinterpreting tasks. Existing solutions address this gap, prompting AI to communicate its understanding in a structured natural-language document. This document can then be inspected, edited, and approved by the developer. While effective, the developer still needs to vet the resulting AI-generated code for safety and reliability, requiring both domain and coding expertise. Our goal is to decouple this requirement, paving the way for numerous organizations and individuals, including those without coding expertise, to develop software. 

PwR approach

Programming with Representations (PwR, pronounced “power”), which we are presenting at the Agami Summit 2023 (opens in new tab), is a software development approach that relies on a domain-specific language (DSL), or representation, defined by a developer specializing in a specific domain. This representation includes built-in guardrails that are automatically implemented throughout the software development process. Once a representation is defined for a domain, PwR enables any developer interested in that domain to translate their intentions using natural language into a program in that representation. This process is illustrated in Figure 1. 

PwR uses large language models (LLMs) to interpret user conversations and transform them into DSL programs. This process involves traversing a code-generation pipeline to ultimately derive executable code. However, despite advancements in LLMs’ code generation, these models still grapple with limitations like hallucinations and limited context windows. Using DSL reduces the amount of code that LLMs need to generate, as most code can be generated from the DSL, increasing accuracy throughout the process.

The DSL incorporates guardrails, ensuring that the essential components are there, such as the starting state of a workflow, clearly defined transitions, and error handling protocols. These guardrails can be automatically examined and communicated back to the developer in natural language, allowing for necessary corrections. While certain guardrails may enhance safety preferences, developers still must confirm that the intended functionality was implemented. PwR simply acts as a facilitator within the cycle involving the developer, the LLM, and the DSL checker. 

PwR does not require developers to learn a custom DSL. Instead, it generates a natural-language representation (NLR) of the DSL. Developers can inspect this NLR, essentially programming in a natural language representation while the underlying DSL remains concealed. This approach grants developers the flexibility and ease of interacting with a natural-language representation while preserving the precision of their intent within the DSL. Additionally, developers can access a live test environment where their code can be hosted and tested for functionality. These capabilities are integrated into the PwR Studio tool, making it easy to get started.

PwR lowers the programming barrier, empowering nontechnical domain experts like teachers to create software tailored to their specific needs. Additionally, it can improve productivity for complex, multidisciplinary software engineering teams, enabling them to efficiently handle large volumes of changes.

Creating a welfare scheme application with PwR

Let’s take an example of how PwR can be applied. In a scenario where a nongovernment organization (NGO) aims to develop an application facilitating citizen access to government welfare schemes—enabling search, identification, and application processes involving authentication and deposits—the orchestration of multiple components is crucial. It is vital to accurately set up these components before deploying them at scale, given the program’s involvement with user data and monetary transactions. 

Reliable orchestration of these types of components can significantly enhance all types applications. We initiate this process by building a custom DSL, encoding interconnected workflows comprising various tasks. Each task represents a singular action that might involve calling an external API or another workflow. This DSL seamlessly interacts with external APIs through plugins available through the PwR Studio store. 

The following video demonstrates how PwR Studio, configured with the DSL workflow, constructs the NGO application. A developer augments the initial version of the application by incorporating the payment feature. This is accomplished by conversing with PwR Studio, understanding the specific requirements, and implementing necessary modifications. Additionally, the developer gains access to a test environment where they can launch and interact with the application in a controlled setting. 

Looking forward

We intend to provide PwR Studio as an open-source integrated development environment (IDE) for creating software through conversations. Our initial aim is to facilitate workflow-based applications for NGOs and social enterprises that have little access to technical expertise. However, our ambitions stretch far beyond this scope.

With the recent success of GitHub Copilot for conversational code generation and recent announcements surrounding OpenAI’s GPTs framework for programming ChatGPT-like bots, it’s evident that AI is poised to democratize software development, granting everyone the ability to create software. With that, it’s imperative to prioritize safety and reliability. PwR is an approach that incorporates these priorities, where the insights of a few technical experts guide a large community of developers through the power of representation. We encourage the software development community to experiment with PwR, and similar ideas, to build safe and reliable AI-powered software.

Learn more on the PwR project page.

Acknowledgements

PwR is the result of a joined collaboration with several of our colleagues, including Sriram Rajamani, B. Ashok, Mohit Jain, Vageesh D C, Dinesh KA, and Sanoop Menon. We would also like to thank Vyshak Jain, Drishti Goel, Hamna, and Sanoop Menon for their help in creating the video.

Opens in a new tab

The post PwR: Using representations for AI-powered software development appeared first on Microsoft Research.

Read More