Armando Solar-Lezama

Massachusetts Institute of Technology | MIT · Department of Electrical Engineering and Computer Science

We present MathDSL, a Domain-Specific Language (DSL) for mathematical equation solving, which, when deployed in program synthesis models, outperforms state-of-the-art reinforcement-learning-based methods. We also introduce a quantitative metric for measuring the conciseness of a mathematical solution and demonstrate the improvement in the quality o...

Skills are temporal abstractions that are intended to improve reinforcement learning (RL) performance through hierarchical RL. Despite our intuition about the properties of an environment that make skills useful, a precise characterization has been absent. We provide the first such characterization, focusing on the utility of deterministic skills i...

Robots are often built from standardized assemblies, (e.g. arms, legs, or fingers), but each robot must be trained from scratch to control all the actuators of all the parts together. In this paper we demonstrate a new approach that takes a single robot and its controller as input and produces a set of modular controllers for each of these assembli...

Sequence-specific RNA-binding proteins (RBPs) play central roles in splicing decisions. Here, we describe a modular splicing architecture that leverages in vitro-derived RNA affinity models for 79 human RBPs and the annotated human genome to produce improved models of RBP binding and activity. Binding and activity are modeled by separate Motif and...

We present a new general-purpose synthesis technique for generating programs from input-output examples. Our method, called metric program synthesis, relaxes the observational equivalence idea (used widely in bottom-up enumerative synthesis) into a weaker notion of observational similarity, with the goal of reducing the search space that the synthe...

Sequence-specific RNA-binding proteins (RBPs) play central roles in splicing decisions, but their exact binding locations and activities are difficult to predict. Here, we describe a modular splicing architecture that leverages in vitro -derived RNA affinity models for 79 human RBPs and the annotated human genome to produce improved models of RBP b...

We present a new dataset for learning to solve, explain, and generate university-level STEM questions from 27 courses across a dozen departments in seven universities. We scale up previous approaches to questions from courses in the departments of Mechanical Engineering, Materials Science and Engineering, Chemistry, Electrical Engineering, Computer...

Large Language Models (LLMs) have shown remarkable aptitude in code generation but still struggle on challenging programming tasks. Self-repair -- in which the model debugs and fixes mistakes in its own code -- has recently become a popular way to boost performance in these settings. However, only very limited studies on how and when self-repair wo...

We curate a comprehensive dataset of 4,550 questions and solutions from problem sets, midterm exams, and final exams across all MIT Mathematics and Electrical Engineering and Computer Science (EECS) courses required for obtaining a degree. We evaluate the ability of large language models to fulfill the graduation requirements for any MIT major in M...

Expert problem-solving is driven by powerful languages for thinking about problems and their solutions. Acquiring expertise means learning these languages—systems of concepts, alongside the skills to use them. We present DreamCoder, a system that learns to solve problems by writing programs. It builds expertise by creating domain-specific programmi...

Real-world processes often contain intermediate state that can be modeled as an extremely sparse tensor. We introduce Sparling, a new kind of informational bottleneck that explicitly models this state by enforcing extreme activation sparsity. We additionally demonstrate that this technique can be used to learn the true intermediate representation w...

We present a new algorithm that synthesizes functional reactive programs from observation data. The key novelty is to iterate between a functional synthesis step, which attempts to generate a transition function over observed states, and an automata synthesis step, which adds any additional latent state necessary to fully account for the observatio...

This paper introduces corpus-guided top-down synthesis as a mechanism for synthesizing library functions that capture common functionality from a corpus of programs in a domain specific language (DSL). The algorithm builds abstractions directly from initial DSL primitives, using syntactic pattern matching of intermediate abstractions to intelligent...

We provide a new multi-task benchmark for evaluating text-to-image models. We perform a human evaluation comparing the most common open-source (Stable Diffusion) and commercial (DALL-E 2) models. Twenty computer science AI graduate students evaluated the two models, on three tasks, at three difficulty levels, across ten prompts each, providing 3,60...

Humans tame the complexity of mathematical reasoning by developing hierarchies of abstractions. With proper abstractions, solutions to hard problems can be expressed concisely, thus making them more likely to be found. In this paper, we propose Learning Mathematical Abstractions (LEMMA): an algorithm that implements this idea for reinforcement lear...

We introduce ObSynth, an interactive system leveraging the domain knowledge embedded in large language models (LLMs) to help users design object models from high level natural language prompts. This is an example of specification reification, the process of taking a high-level, potentially vague specification and reifying it into a more concrete fo...

Neurosymbolic Programming (NP) techniques have the potential to accelerate scientific discovery across fields. These models combine neural and symbolic components to learn complex patterns and representations from data, using high-level concepts or known constraints. As a result, NP techniques can interface with symbolic domain knowledge from scien...

Neurosymbolic Programming (NP) techniques have the potential to accelerate scientific discovery across fields. These models combine neural and symbolic components to learn complex patterns and representations from data, using high-level concepts or known constraints. As a result, NP techniques can interface with symbolic domain knowledge from scien...

Automated, data-driven construction and evaluation of scientific models and theories is a long-standing challenge in artificial intelligence. We present a framework for algorithmically synthesizing models of a basic part of human language: morpho-phonology, the system that builds word forms from sounds. We integrate Bayesian inference with program...

We develop the first theory of control-flow graphs from first principles, and use it to create an algorithm for automatically synthesizing many variants of control-flow graph generators from a language’s operational semantics. Our approach first introduces a new algorithm for converting a large class of small-step operational semantics to an abstra...

Many problem domains, including program synthesis and rewrite-based optimization, require searching astronomically large spaces of programs. Existing approaches often rely on building specialized data structures—version-space algebras, finite tree automata, or e-graphs—to compactly represent such spaces. At their core, all these data structures exp...

Many problem domains, including program synthesis and rewrite-based optimization, require searching astronomically large spaces of programs. Existing approaches often rely on building specialized data structures -- version-space algebras, finite tree automata, or e-graphs -- to compactly represent these programs. To find a compact representation, e...

In this paper, we present a new method for solving the inverse constructive solid geometry (CSG) problem, which aims to generate a programmatic representation of a scene from its unstructured representation (e.g., a raster image). Our method, called metric program synthesis, hinges on the observation that many programs in the inverse CSG domain pro...

Reinforcement learning is a promising strategy for automatically training policies for challenging control tasks. However, state-of-the-art deep reinforcement learning algorithms focus on training deep neural network (DNN) policies, which are black box models that are hard to interpret and reason about. In this chapter, we describe recent progress...

We survey recent work on neurosymbolic programming, an emerging area that bridges the areas of deep learning and program synthesis. Like in classic machine learning, the goal here is to learn functions from data. However, these functions are represented as programs that can use neural modules in addition to symbolic primitives and are induced using...

Physical products are often complex assemblies combining a multitude of 3D parts modeled in computer-aided design (CAD) software. CAD designers build up these assemblies by aligning individual parts to one another using constraints called joints. In this paper we introduce JoinABLe, a learning-based method that assembles parts together to form join...

Ensuring safety for human-interactive robotics is important due to the potential for human injury. The key challenge is defining safety in a way that accounts for the complex range of human behaviors without modeling the human as an unconstrained adversary. We propose a novel approach to ensuring safety in these settings. Our approach focuses on de...

Parametric computer-aided design (CAD) is a standard paradigm used to design manufactured objects, where a 3D shape is represented as a program supported by the CAD software. Despite the pervasiveness of parametric CAD and a growing interest from the research community, currently there does not exist a dataset of realistic CAD models in a concise p...

Sketch is a popular program synthesis tool that solves for unknowns in a sketch or partial program. However, while Sketch is powerful, it does not directly support modular synthesis of dependencies, potentially limiting scalability. In this paper, we introduce Sketcham, a new technique that modularizes a regular sketch by automatically generating m...

A key challenge for reinforcement learning is solving long-horizon planning and control problems. Recent work has proposed leveraging programs to help guide the learning algorithm in these settings. However, these approaches impose a high manual burden on the user since they must provide a guiding program for every new task they seek to achieve. We...

We study the problem of inferring communication structures that can solve cooperative multi-agent planning problems while minimizing the amount of communication. We quantify the amount of communication as the maximum degree of the communication graph; this metric captures settings where agents have limited bandwidth. Minimizing communication is cha...

Neurosymbolic programming is an emerging area that bridges the areas of deep learning and program synthesis. As in classical machine learning, the goal is to learn functions from data. However, these functions are represented as programs that can use neural modules in addition to symbolic primitives and are induced using a combination of symbolic s...

Sketch is a popular program synthesis tool that solves for unknowns in a sketch or partial program. However, while Sketch is powerful, it does not directly support modular synthesis of dependencies, potentially limiting scalability. In this paper, we introduce Sketcham, a new technique that modularizes a regular sketch by automatically generating m...

Synthesizing programs from examples requires searching over a vast, combinatorial space of possible programs. In this search process, a key challenge is representing the behavior of a partially written program before it can be executed, to judge if it is on the right track and predict where to search next. We introduce a general technique for repre...

2020 Owner/Author. Optimizing the physical data storage and retrieval of data are two key database management problems. In this paper, we propose a language that can express both a relational query and the layout of its data. Our language can express a wide range of physical database layouts, going well beyond the row- and column-based methods that...

We develop the first theory of control-flow graphs from first principles, and use it to create an algorithm for automatically synthesizing many variants of control-flow graph generators from a language's operational semantics. Our approach first introduces a new algorithm for converting a large class of small-step operational semantics to an abstra...

Parametric computer-aided design (CAD) is a standard paradigm used for the design of manufactured objects. CAD designers perform modeling operations, such as sketch and extrude, to form a construction sequence that makes up a final design. Despite the pervasiveness of parametric CAD and growing interest from the research community, a dataset of hum...

We present Lifty, a domain-specific language for data-centric applications that manipulate sensitive data. A Lifty programmer annotates the sources of sensitive data with declarative security policies, and the language statically and automatically verifies that the application handles the data according to the policies. Moreover, if verification fa...

Program synthesis techniques construct or infer programs from user-provided specifications, such as input-output examples. Yet most specifications, especially those given by end-users, leave the synthesis problem radically ill-posed, because many programs may simultaneously satisfy the specification. Prior work resolves this ambiguity by using vari...

The shortage of people trained in STEM fields is becoming acute, and universities and colleges are straining to satisfy this demand. In the case of computer science, for instance, the number of US students taking introductory courses has grown three-fold in the past decade. Recently, massive open online courses (MOOCs) have been promoted as a way t...

Deploying deep reinforcement learning in safety-critical settings requires developing algorithms that obey hard constraints during exploration. This paper contributes a first approach toward enforcing formal safety constraints on end-to-end policies with visual inputs. Our approach draws on recent advances in object detection and automated reasonin...

Expert problem-solving is driven by powerful languages for thinking about problems and their solutions. Acquiring expertise means learning these languages -- systems of concepts, alongside the skills to use them. We present DreamCoder, a system that learns to solve problems by writing programs. It builds expertise by creating programming languages...

2020 Owner/Author. We present a new approach to semantic code search based on equational reasoning, and the Yogo tool implementing this approach. Our approach works by considering not only the dataflow graph of a function, but also the dataflow graphs of all equivalent functions reachable via a set of rewrite rules. In doing so, it can recognize an...

Many aspects of human reasoning, including language, require learning rules from very little data. Humans can do this, often learning systematic rules from very few examples, and combining these rules to form compositional rule-based systems. Current neural architectures, on the other hand, often fail to generalize in a compositional manner, especi...

We propose a novel approach to program synthesis, focusing on synthesizing database queries. At a high level, our proposed algorithm takes as input a sketch with soft constraints encoding user intent, and then iteratively interacts with the user to refine the sketch. At each step, our algorithm proposes a candidate refinement of the sketch, which t...

As machine learning systems are increasingly used to make real world legal and financial decisions, it is of paramount importance that we develop algorithms to verify that these systems do not discriminate against minorities. We design a scalable algorithm for verifying fairness specifications. Our algorithm obtains strong correctness guarantees ba...

A key challenge in program synthesis is synthesizing programs that use libraries, which most real-world software does. The current state of the art is to model libraries with mock library implementations that perform the same function in a simpler way. However, mocks may still be large and complex, and must include many implementation details, both...

We formulate numerically-robust inductive proof rules for unbounded stability and safety properties of continuous dynamical systems. These induction rules robustify standard notions of Lyapunov functions and barrier certificates so that they can tolerate small numerical errors. In this way, numerically-driven decision procedures can establish a sou...

We present a neural program synthesis approach integrating components which write, execute, and assess code to navigate the search space of possible programs. We equip the search process with an interpreter or a read-eval-print-loop (REPL), which immediately executes partially written programs, exposing their semantics. The REPL addresses a basic c...

The need to condition distributional properties such as expectation, variance, and entropy arises in algorithmic fairness, model simplification, robustness and many other areas. At face value however, distributional properties are not random variables, and hence conditioning them is a semantic error and type error in probabilistic programming langu...

Optimizing the physical data storage and retrieval of data are two key database management problems. In this paper, we propose a language that can express a wide range of physical database layouts, going well beyond the row- and column- based methods that are widely used in database management systems. We also build a compiler for this language, wh...

Our goal is to build systems which write code automatically from the kinds of specifications humans can most easily provide, such as examples and natural language instruction. The key idea of this work is that a flexible combination of pattern recognition and explicit reasoning can be used to solve these complex programming problems. We propose a m...

We develop a likelihood free inference procedure for conditioning a probabilistic model on a predicate. A predicate is a Boolean valued function which expresses a yes/no question about a domain. Our contribution, which we call predicate exchange, constructs a softened predicate which takes value in the unit interval [0, 1] as opposed to a simply tr...

2019 Neural information processing systems foundation. All rights reserved. We present a neural program synthesis approach integrating components which write, execute, and assess code to navigate the search space of possible programs. We equip the search process with an interpreter or a read-eval-print-loop (REPL), which immediately executes partia...

While computer-aided design is a major part of many modern manufacturing pipelines, the design files typically generated describe raw geometry. Lost in this representation is the procedure by which these designs were generated. In this paper, we present a method for reverse-engineering the process by which 3D models may have been generated, in the...

As machine learning systems are increasingly used to make real world legal and financial decisions, it is of paramount importance that we develop algorithms to verify that these systems do not discriminate against minorities. We design a scalable algorithm for verifying fairness specifications. Our algorithm obtains strong correctness guarantees ba...

Writing programs that are both correct and efficient is challenging. A potential solution lies in program synthesis aimed at automatic derivation of an executable implementation (the “how”) from a high-level logical specification of the desired input-to-output behavior (the “what”). A mature synthesis technology can have a transformative impact on...

We present a new approach for building source-to-source transformations that can run on multiple programming languages, based on a new way of representing programs called incremental parametric syntax. We implement this approach in Haskell in our Cubix system, and construct incremental parametric syntaxes for C, Java, JavaScript, Lua, and Python. W...

Delimited continuations are the mother of all monads! So goes the slogan inspired by Filinski’s 1994 paper, which showed that delimited continuations can implement any monadic effect, letting the programmer use an effect as easily as if it was built into the language. It’s a shame that not many languages have delimited continuations. Luckily, excep...

Solving nonlinear SMT problems over real numbers has wide applications in robotics and AI. While significant progress is made in solving quantifier-free SMT formulas in the domain, quantified formulas have been much less investigated. We propose the first delta-complete algorithm for solving satisfiability of nonlinear SMT over real numbers with un...

We propose \(\delta \)-complete decision procedures for solving satisfiability of nonlinear SMT problems over real numbers that contain universal quantification and a wide range of nonlinear functions. The methods combine interval constraint propagation, counterexample-guided synthesis, and numerical optimization. In particular, we show how to hand...

In this position paper, we describe our vision of the future of machine programming through a categorical examination of three pillars of research. Those pillars are: (i) intention, (ii) invention, and (iii) adaptation. Intention emphasizes advancements in the human-to-computer and computer-to-machine-learning interfaces. Invention emphasizes the c...

In this position paper, we describe our vision of the future of machine programming through a categorical examination of three pillars of research. Those pillars are: (i) intention, (ii) invention, and (iii) adaptation. Intention emphasizes advancements in the human-to-computer and computer-to-machine-learning interfaces. Invention emphasizes the c...

While deep reinforcement learning has successfully solved many challenging control tasks, its real-world applicability has been limited by the inability to ensure the safety of learned policies. We propose an approach to verifiable reinforcement learning by training decision tree policies, which can represent complex policies (since they are nonpar...

In this position paper, we describe our vision of the future of machine-based programming through a categorical examination of three pillars of research. Those pillars are: (i) intention, (ii) invention, and(iii) adaptation. Intention emphasizes advancements in the human-to-computer and computer-to-machine-learning interfaces. Invention emphasizes...

The paper describes a new algorithm to generate minimal, stable, and symbolic corrections to an input that will cause a neural network with ReLU neurons to change its output. We argue that such a correction is a useful way to provide feedback to a user when the neural network produces an output that is different from a desired output. Our algorithm...

In this paper, we present ReaS, a technique that combines numerical optimization with SAT solving to synthesize unknowns in a program that involves discrete and floating point computation. ReaS makes the program end-to-end differentiable by smoothing any Boolean expression that introduces discontinuity such as conditionals and relaxing the Boolean...

Successful approaches to program induction require a hand-engineered domain-specific language (DSL), constraining the space of allowed programs and imparting prior knowledge of the domain. We contribute a program induction algorithm called EC 2 that learns a DSL while jointly training a neural network to efficiently search for programs in the learn...

2018 Curran Associates Inc.All rights reserved. While deep reinforcement learning has successfully solved many challenging control tasks, its real-world applicability has been limited by the inability to ensure the safety of learned policies. We propose an approach to verifiable reinforcement learning by training decision tree policies, which can r...