While the value of artificial intelligence (AI) is just beginning to emerge in the enterprise, some 46% of CIOs have plans to implement the technology in the future, according to a survey from research firm Gartner. "Despite huge levels of interest in AI technologies, current implementations remain at quite low levels," Gartner analyst Whit Andrews wrote in a press release. "However, there is potential for strong growth as CIOs begin piloting AI programmes through a combination of buy, build and outsource efforts." Early AI projects have demonstrated the value of the technology, but have also showed some of the challenges linked to its adoption, the release said. Here are four of the main lessons that the enterprise has learned from early AI initiatives.

Thank you for your help building out the list!]

Deep learning methods are useful for high-dimensional data and are becoming widely used in many areas of software engineering. Deep learners utilizes extensive computational power and can take a long time to train-- making it difficult to widely validate and repeat and improve their results. Further, they are not the best solution in all domains. For example, recent results show that for finding related Stack Overflow posts, a tuned SVM performs similarly to a deep learner, but is significantly faster to train. This paper extends that recent result by clustering the dataset, then tuning very learners within each cluster. This approach is over 500 times faster than deep learning (and over 900 times faster if we use all the cores on a standard laptop computer). Significantly, this faster approach generates classifiers nearly as good (within 2\% F1 Score) as the much slower deep learning method. Hence we recommend this faster methods since it is much easier to reproduce and utilizes far fewer CPU resources. More generally, we recommend that before researchers release research results, that they compare their supposedly sophisticated methods against simpler alternatives (e.g applying simpler learners to build local models).

Fortunately, in many cases such data are essentially low dimensional; i.e., they stay on a low-dimensional manifold embedded in the high dimensional space [1]. This key observation suggests the possibility of dimensionality reduction to facilitate visualization and analysis of the data. Manifold learning is one of the mainstream nonlinear dimensionality reduction techniques [2]. Driven by major academic and practical motives, many algorithms [3-9] were developed to flatten an embedded manifold and reveal an intrinsic structure. One of the representative methods, Isomap combines the Floyd-Warshall algorithm with multidimensional scaling (MDS [10]) to compress high-dimensional data.

Whether it is a minor tweak, or a major new update, releasing a new version of a running system is a stressful time. While the release has typically gone through rounds of offline testing, real world testing often uncovers additional corner cases that may manifest themselves as bugs, inefficiencies, or overall poor performance. This is especially the case in machine learning applications, where models are typically trained to maximize a proxy objective, and a model that performs better on offline metrics is not guaranteed to work well in practice. The usual approach in such scenarios is to evaluate the new system through a series of closely monitored A/B tests. The new version is usually released to a small number of customers, and, if no concerns are found and metrics look good, the portion of traffic served by the new system is slowly increased. While A/B tests provide a sense of safety in that a detrimental change will be quickly observed and corrected (or rolled back), they are not a silver bullet. First, A/B tests are labor intensive--they are typically monitored manually, with an engineer, or a technician, checking the results of the test on a regular basis (for example, daily or weekly). Second, the evaluation is usually dependent on average metrics--e.g.

Learning to learn is a powerful paradigm for enabling models to learn from data more effectively and efficiently. A popular approach to meta-learning is to train a recurrent model to read in a training dataset as input and output the parameters of a learned model, or output predictions for new test inputs. Alternatively, a more recent approach to meta-learning aims to acquire deep representations that can be effectively fine-tuned, via standard gradient descent, to new tasks. In this paper, we consider the meta-learning problem from the perspective of universality, formalizing the notion of learning algorithm approximation and comparing the expressive power of the aforementioned recurrent models to the more recent approaches that embed gradient descent into the meta-learner. In particular, we seek to answer the following question: does deep representation combined with standard gradient descent have sufficient capacity to approximate any learning algorithm? We find that this is indeed true, and further find, in our experiments, that gradient-based meta-learning consistently leads to learning strategies that generalize more widely compared to those represented by recurrent models.

Hyperparameter tuning is one of the most time-consuming workloads in deep learning. State-of-the-art optimizers, such as AdaGrad, RMSProp and Adam, reduce this labor by adaptively tuning an individual learning rate for each variable. Recently researchers have shown renewed interest in simpler methods like momentum SGD as they may yield better test metrics. Motivated by this trend, we ask: can simple adaptive methods based on SGD perform as well or better? We revisit the momentum SGD algorithm and show that hand-tuning a single learning rate and momentum makes it competitive with Adam. We then analyze its robustness to learning rate misspecification and objective curvature variation. Based on these insights, we design YellowFin, an automatic tuner for momentum and learning rate in SGD. YellowFin optionally uses a negative-feedback loop to compensate for the momentum dynamics in asynchronous settings on the fly. We empirically show that YellowFin can converge in fewer iterations than Adam on ResNets and LSTMs for image recognition, language modeling and constituency parsing, with a speedup of up to 3.28x in synchronous and up to 2.69x in asynchronous settings.

Crowdsourcing is an important avenue for collecting machine learning data, but crowdsourcing can go beyond simple data collection by employing the creativity and wisdom of crowd workers. Yet crowd participants are unlikely to be experts in statistics or predictive modeling, and it is not clear how well non-experts can contribute creatively to the process of machine learning. Here we study an end-to-end crowdsourcing algorithm where groups of non-expert workers propose supervised learning problems, rank and categorize those problems, and then provide data to train predictive models on those problems. Problem proposal includes and extends feature engineering because workers propose the entire problem, not only the input features but also the target variable. We show that workers without machine learning experience can collectively construct useful datasets and that predictive models can be learned on these datasets. In our experiments, the problems proposed by workers covered a broad range of topics, from politics and current events to problems capturing health behavior, demographics, and more. Workers also favored questions showing positively correlated relationships, which has interesting implications given many supervised learning methods perform as well with strong negative correlations. Proper instructions are crucial for non-experts, so we also conducted a randomized trial to understand how different instructions may influence the types of problems proposed by workers. In general, shifting the focus of machine learning tasks from designing and training individual predictive models to problem proposal allows crowdsourcers to design requirements for problems of interest and then guide workers towards contributing to the most suitable problems.

Catastrophic forgetting occurs when a neural network loses the information learned in a previous task after training on subsequent tasks. This problem remains a hurdle for artificial intelligence systems with sequential learning capabilities. In this paper, we propose a task-based hard attention mechanism that preserves previous tasks' information without affecting the current task's learning. A hard attention mask is learned concurrently to every task, through stochastic gradient descent, and previous masks are exploited to condition such learning. We show that the proposed mechanism is effective for reducing catastrophic forgetting, cutting current rates by 45 to 80%. We also show that it is robust to different hyperparameter choices, and that it offers a number of monitoring capabilities. The approach features the possibility to control both the stability and compactness of the learned knowledge, which we believe makes it also attractive for online learning or network compression applications.

Teachers intentionally pick the most informative examples to show their students. However, if the teacher and student are neural networks, the examples that the teacher network learns to give, although effective at teaching the student, are typically uninterpretable. We show that training the student and teacher iteratively, rather than jointly, can produce interpretable teaching strategies. We evaluate interpretability by (1) measuring the similarity of the teacher's emergent strategies to intuitive strategies in each domain and (2) conducting human experiments to evaluate how effective the teacher's strategies are at teaching humans. We show that the teacher network learns to select or generate interpretable, pedagogical examples to teach rule-based, probabilistic, boolean, and hierarchical concepts.