H Company, also known simply as H, is a French artificial intelligence startup which develops "action-oriented" artificial intelligence agents for enterprise automation and productivity. In May 2024, H Company closed a record-setting $220 million seed round, at the time the largest AI raise in Europe. In 2026, H Company released Holo 3, the latest generation of its computer-use AI models. The update marked a major advance in agentic AI, enabling agents to navigate any user interface, interpret screens, and complete complex, multi-step tasks across enterprise systems—much like a human user. This breakthrough positioned H Company at the frontier of computer-use autonomy, accelerating the integration of AI in enterprise workflows. == History == H Company was founded in 2023 in Paris by Laurent Sifre, Charles Kantor, and three DeepMind veterans: Daan Wiestra, Karl Tuyls, Julien Perollat. In May 2024, the firm secured what was then the largest European AI seed round, totaling $220 million led by US investors including Eric Schmidt (former Google CEO), Amazon, and backed by Accel, Bpifrance, UiPath, Eurazeo, Xavier Niel, Yuri Milner, Bernard Arnault, Samsung and others. In August 2024, three cofounders (Wiestra, Tuyls, Perollat) left the company over operational disagreements. In November 2024, H launched Runner H, its first agentic-API platform, which combined a large language model (LLM) and a reduced, 2-billion parameter vision-language model (VLM). In May 2025, H Company acquired Mithril Security, and in June 2025 the company widened its offering for agentic models. In June 2025, Gautier Cloix (formerly CEO Palantir France) replaced Charles Kantor as CEO of H Company, aiming to pivot the company towards a "forward deployed engineers" model. In July 2025, H Company introduced Surfer-H-CLI, an open-source, web-native Chrome agent designed for browser-based automation—able to search, scroll, click, and type on behalf of users and controllable via any visual language model (VLM). When paired with its June 2025 open-sourced 3B-parameter Holo-1 model, Surfer-H-CLI achieved 92.2% WebVoyager benchmark accuracy. == Activity == H Company creates enterprise AI models and agents (agentic AI) to automate and optimize complex workflows. H Company specifically designs AI agents called computer use capable of autonomously interfacing with any software (local or cloud-based) to detect and automate repetitive operations. H Company is based in Paris, France, with international offices in London and New York. H Company raised $220 million since its inception. Gautier Cloix is president and CEO of the company. H Company client include the French national lottery FDJ United. In March 2026, H Company released Holo3, a family of artificial intelligence models designed to operate digital systems by interacting directly with user interfaces. Holo3 enables agents ("virtual humanoids") to understand what is displayed in front-end environments—such as web pages, desktop applications, and other graphical user interfaces—and perform actions such as clicking, typing, and navigating across them to complete multi-step tasks. On the OSWorld-Verified benchmark, Holo3 reportedly achieved about 78.9%, surpassing the scores of OpenAI’s GPT‑5.4 and Anthropic’s Claude Opus 4.6 on this specific test, at roughly one-tenth of the inference cost of these proprietary systems. The release has been presented as a significant step toward automating routine digital workflows, allowing organizations to offload repetitive on-screen work, such as data entry and reconciliation across multiple tools, to AI-based agents.
Conduit (company)
Conduit Ltd. is an international software company. From its founding in 2005 to 2013, its most well-known product was the Conduit toolbar, which was widely-described as malware. In 2013, it spun off its toolbar business; today, its main product is a mobile development platform that allows users to create native and web mobile applications for smartphones. == Products == From 2005 to 2013, the company's most well-known product was the Conduit toolbar, which is flagged by most antivirus software as potentially unwanted and adware. Conduit's toolbar software is often downloaded by malware packages from other publishers. The company spun off the toolbar division that manages the Conduit toolbar in 2013. Today, the company's main product is a mobile development platform that allows users to create native and web mobile applications for smartphones. App creation for its App Gallery is free, but it charges a monthly subscription fee to place apps on the App Store or Google Play. == History == Conduit was founded in 2005 by Shilo, Dror Erez, and Gaby Bilcyzk. Between years 2005 and 2013, it ran a successful but controversial toolbar platform business. Conduit was part of the so-called Download Valley companies monetizing free software and downloads by bundling adware. The toolbars were criticized by some as being very difficult to uninstall. The toolbar software was referred to as a "potentially unwanted program" by some in the computer industry because it could be used to change browser settings. The company had more than 400 employees in 2013. In September same year, Conduit spun off its entire website toolbar business division, which combined with Perion Network. After the deal, Conduit shareholders owned 81% of Perion's existing shares and both Perion and Conduit remained independent companies. The substantial size of the Conduit user base allowed Perion to immediately surpass AOL in U.S. searches. In 2015, Conduit announced it would purchase Keeprz, a mobile customer loyalty platform, for $45 million.
Q-learning
Q-learning is a reinforcement learning algorithm that trains an agent to assign values to its possible actions based on its current state, without requiring a model of the environment (model-free). It can handle problems with stochastic transitions and rewards without requiring adaptations. For example, in a grid maze, an agent learns to reach an exit worth 10 points. At a junction, Q-learning might assign a higher value to moving right than left if right gets to the exit faster, improving this choice by trying both directions over time. For any finite Markov decision process, Q-learning finds an optimal policy in the sense of maximizing the expected value of the total reward over any and all successive steps, starting from the current state. Q-learning can identify an optimal action-selection policy for any given finite Markov decision process, given infinite exploration time and a partly random policy. "Q" refers to the function that the algorithm computes: the expected reward—that is, the quality—of an action taken in a given state. == Reinforcement learning == Reinforcement learning involves an agent, a set of states S {\displaystyle {\mathcal {S}}} , and a set A {\displaystyle {\mathcal {A}}} of actions per state. By performing an action a ∈ A {\displaystyle a\in {\mathcal {A}}} , the agent transitions from state to state. Executing an action in a specific state provides the agent with a reward (a numerical score). The goal of the agent is to maximize its total reward. It does this by adding the maximum reward attainable from future states to the reward for achieving its current state, effectively influencing the current action by the potential future reward. This potential reward is a weighted sum of expected values of the rewards of all future steps starting from the current state. As an example, consider the process of boarding a train, in which the reward is measured by the negative of the total time spent boarding (alternatively, the cost of boarding the train is equal to the boarding time). One strategy is to enter the train door as soon as they open, minimizing the initial wait time for yourself. If the train is crowded, however, then you will have a slow entry after the initial action of entering the door as people are fighting you to depart the train as you attempt to board. The total boarding time, or cost, is then: 0 seconds wait time + 15 seconds fight time On the next day, by random chance (exploration), you decide to wait and let other people depart first. This initially results in a longer wait time. However, less time is spent fighting the departing passengers. Overall, this path has a higher reward than that of the previous day, since the total boarding time is now: 5 second wait time + 0 second fight time Through exploration, despite the initial (patient) action resulting in a larger cost (or negative reward) than in the forceful strategy, the overall cost is lower, thus revealing a more rewarding strategy. == Algorithm == After Δ t {\displaystyle \Delta t} steps into the future the agent will decide some next step. The weight for this step is calculated as γ Δ t {\displaystyle \gamma ^{\Delta t}} , where γ {\displaystyle \gamma } (the discount factor) is a number between 0 and 1 ( 0 ≤ γ ≤ 1 {\displaystyle 0\leq \gamma \leq 1} ). Assuming γ < 1 {\displaystyle \gamma <1} , it has the effect of valuing rewards received earlier higher than those received later (reflecting the value of a "good start"). γ {\displaystyle \gamma } may also be interpreted as the probability to succeed (or survive) at every step Δ t {\displaystyle \Delta t} . The algorithm, therefore, has a function that calculates the quality of a state–action combination: Q : S × A → R {\displaystyle Q:{\mathcal {S}}\times {\mathcal {A}}\to \mathbb {R} } . Before learning begins, Q {\displaystyle Q} is initialized to a possibly arbitrary fixed value (chosen by the programmer). Then, at each time t {\displaystyle t} the agent selects an action A t {\displaystyle A_{t}} , observes a reward R t + 1 {\displaystyle R_{t+1}} , enters a new state S t + 1 {\displaystyle S_{t+1}} (that may depend on both the previous state S t {\displaystyle S_{t}} and the selected action), and Q {\displaystyle Q} is updated. The core of the algorithm is a Bellman equation as a simple value iteration update, using the weighted average of the current value and the new information: Q n e w ( S t , A t ) ← ( 1 − α ⏟ learning rate ) ⋅ Q ( S t , A t ) ⏟ current value + α ⏟ learning rate ⋅ ( R t + 1 ⏟ reward + γ ⏟ discount factor ⋅ max a Q ( S t + 1 , a ) ⏟ estimate of optimal future value ⏟ new value (temporal difference target) ) {\displaystyle Q^{new}(S_{t},A_{t})\leftarrow (1-\underbrace {\alpha } _{\text{learning rate}})\cdot \underbrace {Q(S_{t},A_{t})} _{\text{current value}}+\underbrace {\alpha } _{\text{learning rate}}\cdot {\bigg (}\underbrace {\underbrace {R_{t+1}} _{\text{reward}}+\underbrace {\gamma } _{\text{discount factor}}\cdot \underbrace {\max _{a}Q(S_{t+1},a)} _{\text{estimate of optimal future value}}} _{\text{new value (temporal difference target)}}{\bigg )}} where R t + 1 {\displaystyle R_{t+1}} is the reward received when moving from the state S t {\displaystyle S_{t}} to the state S t + 1 {\displaystyle S_{t+1}} , and α {\displaystyle \alpha } is the learning rate ( 0 < α ≤ 1 ) {\displaystyle (0<\alpha \leq 1)} . Note that Q n e w ( S t , A t ) {\displaystyle Q^{new}(S_{t},A_{t})} is the sum of three terms: ( 1 − α ) Q ( S t , A t ) {\displaystyle (1-\alpha )Q(S_{t},A_{t})} : the current value (weighted by one minus the learning rate) α R t + 1 {\displaystyle \alpha \,R_{t+1}} : the reward R t + 1 {\displaystyle R_{t+1}} to obtain if action A t {\displaystyle A_{t}} is taken when in state S t {\displaystyle S_{t}} (weighted by learning rate) α γ max a Q ( S t + 1 , a ) {\displaystyle \alpha \gamma \max _{a}Q(S_{t+1},a)} : the maximum reward that can be obtained from state S t + 1 {\displaystyle S_{t+1}} (weighted by learning rate and discount factor) An episode of the algorithm ends when state S t + 1 {\displaystyle S_{t+1}} is a final or terminal state. However, Q-learning can also learn in non-episodic tasks (as a result of the property of convergent infinite series). If the discount factor is lower than 1, the action values are finite even if the problem can contain infinite loops or paths. For all final states s f {\displaystyle s_{f}} , Q ( s f , a ) {\displaystyle Q(s_{f},a)} is never updated, but is set to the reward value r {\displaystyle r} observed for state s f {\displaystyle s_{f}} . In most cases, Q ( s f , a ) {\displaystyle Q(s_{f},a)} can be taken to equal zero. == Influence of variables == === Learning rate === The learning rate or step size determines to what extent newly acquired information overrides old information. A factor of 0 makes the agent learn nothing (exclusively exploiting prior knowledge), while a factor of 1 makes the agent consider only the most recent information (ignoring prior knowledge to explore possibilities). In fully deterministic environments, a learning rate of α t = 1 {\displaystyle \alpha _{t}=1} is optimal. When the problem is stochastic, the algorithm converges under some technical conditions on the learning rate that require it to decrease to zero. In practice, often a constant learning rate is used, such as α t = 0.1 {\displaystyle \alpha _{t}=0.1} for all t {\displaystyle t} . === Discount factor === The discount factor γ {\displaystyle \gamma } determines the importance of future rewards. A factor of 0 will make the agent "myopic" (or short-sighted) by only considering current rewards, i.e. r t {\displaystyle r_{t}} (in the update rule above), while a factor approaching 1 will make it strive for a long-term high reward. If the discount factor meets or exceeds 1, the action values may diverge. For γ = 1 {\displaystyle \gamma =1} , without a terminal state, or if the agent never reaches one, all environment histories become infinitely long, and utilities with additive, undiscounted rewards generally become infinite. Even with a discount factor only slightly lower than 1, Q-function learning leads to propagation of errors and instabilities when the value function is approximated with an artificial neural network. In that case, starting with a lower discount factor and increasing it towards its final value accelerates learning. === Initial conditions (Q0) === Since Q-learning is an iterative algorithm, it implicitly assumes an initial condition before the first update occurs. High initial values, also known as "optimistic initial conditions", can encourage exploration: no matter what action is selected, the update rule will cause it to have lower values than the other alternative, thus increasing their choice probability. The first reward r {\displaystyle r} can be used to reset the initial conditions. According to this idea, the first time an action is taken the reward is used to set the value
Policy gradient method
Policy gradient methods are a class of reinforcement learning algorithms and a sub-class of policy optimization methods. Unlike value-based methods which learn a value function to derive a policy, policy optimization methods directly learn a policy function π {\displaystyle \pi } that selects actions without consulting a value function. For policy gradient to apply, the policy function π θ {\displaystyle \pi _{\theta }} is parameterized by a differentiable parameter θ {\displaystyle \theta } . == Overview == In policy-based RL, the actor is a parameterized policy function π θ {\displaystyle \pi _{\theta }} , where θ {\displaystyle \theta } are the parameters of the actor. The actor takes as argument the state of the environment s {\displaystyle s} and produces a probability distribution π θ ( ⋅ ∣ s ) {\displaystyle \pi _{\theta }(\cdot \mid s)} . If the action space is discrete, then ∑ a π θ ( a ∣ s ) = 1 {\displaystyle \sum _{a}\pi _{\theta }(a\mid s)=1} . If the action space is continuous, then ∫ a π θ ( a ∣ s ) d a = 1 {\displaystyle \int _{a}\pi _{\theta }(a\mid s)\mathrm {d} a=1} . The goal of policy optimization is to find some θ {\displaystyle \theta } that maximizes the expected episodic reward J ( θ ) {\displaystyle J(\theta )} : J ( θ ) = E π θ [ ∑ t = 0 T γ t R t | S 0 = s 0 ] {\displaystyle J(\theta )=\mathbb {E} _{\pi _{\theta }}\left[\sum _{t=0}^{T}\gamma ^{t}R_{t}{\Big |}S_{0}=s_{0}\right]} where γ {\displaystyle \gamma } is the discount factor, R t {\displaystyle R_{t}} is the reward at step t {\displaystyle t} , s 0 {\displaystyle s_{0}} is the starting state, and T {\displaystyle T} is the time-horizon (which can be infinite). The policy gradient is defined as ∇ θ J ( θ ) {\displaystyle \nabla _{\theta }J(\theta )} . Different policy gradient methods stochastically estimate the policy gradient in different ways. The goal of any policy gradient method is to iteratively maximize J ( θ ) {\displaystyle J(\theta )} by gradient ascent. Since the key part of any policy gradient method is the stochastic estimation of the policy gradient, they are also studied under the title of "Monte Carlo gradient estimation". == REINFORCE == === Policy gradient === The REINFORCE algorithm, introduced by Ronald J. Williams in 1992, was the first policy gradient method. It is based on the identity for the policy gradient ∇ θ J ( θ ) = E π θ [ ∑ t = 0 T ∇ θ ln π θ ( A t ∣ S t ) ∑ t = 0 T ( γ t R t ) | S 0 = s 0 ] {\displaystyle \nabla _{\theta }J(\theta )=\mathbb {E} _{\pi _{\theta }}\left[\sum _{t=0}^{T}\nabla _{\theta }\ln \pi _{\theta }(A_{t}\mid S_{t})\;\sum _{t=0}^{T}(\gamma ^{t}R_{t}){\Big |}S_{0}=s_{0}\right]} which can be improved via the "causality trick" ∇ θ J ( θ ) = E π θ [ ∑ t = 0 T ∇ θ ln π θ ( A t ∣ S t ) ∑ τ = t T ( γ τ R τ ) | S 0 = s 0 ] {\displaystyle \nabla _{\theta }J(\theta )=\mathbb {E} _{\pi _{\theta }}\left[\sum _{t=0}^{T}\nabla _{\theta }\ln \pi _{\theta }(A_{t}\mid S_{t})\sum _{\tau =t}^{T}(\gamma ^{\tau }R_{\tau }){\Big |}S_{0}=s_{0}\right]} Thus, we have an unbiased estimator of the policy gradient: ∇ θ J ( θ ) ≈ 1 N ∑ n = 1 N [ ∑ t = 0 T ∇ θ ln π θ ( A t , n ∣ S t , n ) ∑ τ = t T ( γ τ − t R τ , n ) ] {\displaystyle \nabla _{\theta }J(\theta )\approx {\frac {1}{N}}\sum _{n=1}^{N}\left[\sum _{t=0}^{T}\nabla _{\theta }\ln \pi _{\theta }(A_{t,n}\mid S_{t,n})\sum _{\tau =t}^{T}(\gamma ^{\tau -t}R_{\tau ,n})\right]} where the index n {\displaystyle n} ranges over N {\displaystyle N} rollout trajectories using the policy π θ {\displaystyle \pi _{\theta }} . The score function ∇ θ ln π θ ( A t ∣ S t ) {\displaystyle \nabla _{\theta }\ln \pi _{\theta }(A_{t}\mid S_{t})} can be interpreted as the direction in the parameter space that increases the probability of taking action A t {\displaystyle A_{t}} in state S t {\displaystyle S_{t}} . The policy gradient, then, is a weighted average of all possible directions to increase the probability of taking any action in any state, but weighted by reward signals, so that if taking a certain action in a certain state is associated with high reward, then that direction would be highly reinforced, and vice versa. === Algorithm === The REINFORCE algorithm is a loop: Rollout N {\displaystyle N} trajectories in the environment, using π θ t {\displaystyle \pi _{\theta _{t}}} as the policy function. Compute the policy gradient estimation: g i ← 1 N ∑ n = 1 N [ ∑ t = 0 T ∇ θ t ln π θ ( A t , n ∣ S t , n ) ∑ τ = t T ( γ τ R τ , n ) ] {\displaystyle g_{i}\leftarrow {\frac {1}{N}}\sum _{n=1}^{N}\left[\sum _{t=0}^{T}\nabla _{\theta _{t}}\ln \pi _{\theta }(A_{t,n}\mid S_{t,n})\sum _{\tau =t}^{T}(\gamma ^{\tau }R_{\tau ,n})\right]} Update the policy by gradient ascent: θ i + 1 ← θ i + α i g i {\displaystyle \theta _{i+1}\leftarrow \theta _{i}+\alpha _{i}g_{i}} Here, α i {\displaystyle \alpha _{i}} is the learning rate at update step i {\displaystyle i} . == Variance reduction == REINFORCE is an on-policy algorithm, meaning that the trajectories used for the update must be sampled from the current policy π θ {\displaystyle \pi _{\theta }} . This can lead to high variance in the updates, as the returns R ( τ ) {\displaystyle R(\tau )} can vary significantly between trajectories. Many variants of REINFORCE have been introduced, under the title of variance reduction. === REINFORCE with baseline === A common way for reducing variance is the REINFORCE with baseline algorithm, based on the following identity: ∇ θ J ( θ ) = E π θ [ ∑ t = 0 T ∇ θ ln π θ ( A t | S t ) ( ∑ τ = t T ( γ τ R τ ) − b ( S t ) ) | S 0 = s 0 ] {\displaystyle \nabla _{\theta }J(\theta )=\mathbb {E} _{\pi _{\theta }}\left[\sum _{t=0}^{T}\nabla _{\theta }\ln \pi _{\theta }(A_{t}|S_{t})\left(\sum _{\tau =t}^{T}(\gamma ^{\tau }R_{\tau })-b(S_{t})\right){\Big |}S_{0}=s_{0}\right]} for any function b : States → R {\displaystyle b:{\text{States}}\to \mathbb {R} } . This can be proven by applying the previous lemma. The algorithm uses the modified gradient estimator g i ← 1 N ∑ n = 1 N [ ∑ t = 0 T ∇ θ t ln π θ ( A t , n | S t , n ) ( ∑ τ = t T ( γ τ R τ , n ) − b i ( S t , n ) ) ] {\displaystyle g_{i}\leftarrow {\frac {1}{N}}\sum _{n=1}^{N}\left[\sum _{t=0}^{T}\nabla _{\theta _{t}}\ln \pi _{\theta }(A_{t,n}|S_{t,n})\left(\sum _{\tau =t}^{T}(\gamma ^{\tau }R_{\tau ,n})-b_{i}(S_{t,n})\right)\right]} and the original REINFORCE algorithm is the special case where b i ≡ 0 {\displaystyle b_{i}\equiv 0} . === Actor-critic methods === If b i {\textstyle b_{i}} is chosen well, such that b i ( S t ) ≈ ∑ τ = t T ( γ τ R τ ) = γ t V π θ i ( S t ) {\textstyle b_{i}(S_{t})\approx \sum _{\tau =t}^{T}(\gamma ^{\tau }R_{\tau })=\gamma ^{t}V^{\pi _{\theta _{i}}}(S_{t})} , this could significantly decrease variance in the gradient estimation. That is, the baseline should be as close to the value function V π θ i ( S t ) {\displaystyle V^{\pi _{\theta _{i}}}(S_{t})} as possible, approaching the ideal of: ∇ θ J ( θ ) = E π θ [ ∑ t = 0 T ∇ θ ln π θ ( A t | S t ) ( ∑ τ = t T ( γ τ R τ ) − γ t V π θ ( S t ) ) | S 0 = s 0 ] {\displaystyle \nabla _{\theta }J(\theta )=\mathbb {E} _{\pi _{\theta }}\left[\sum _{t=0}^{T}\nabla _{\theta }\ln \pi _{\theta }(A_{t}|S_{t})\left(\sum _{\tau =t}^{T}(\gamma ^{\tau }R_{\tau })-\gamma ^{t}V^{\pi _{\theta }}(S_{t})\right){\Big |}S_{0}=s_{0}\right]} Note that, as the policy π θ t {\displaystyle \pi _{\theta _{t}}} updates, the value function V π θ i ( S t ) {\displaystyle V^{\pi _{\theta _{i}}}(S_{t})} updates as well, so the baseline should also be updated. One common approach is to train a separate function that estimates the value function, and use that as the baseline. This is one of the actor-critic methods, where the policy function is the actor and the value function is the critic. The Q-function Q π {\displaystyle Q^{\pi }} can also be used as the critic, since ∇ θ J ( θ ) = E π θ [ ∑ 0 ≤ t ≤ T γ t ∇ θ ln π θ ( A t | S t ) ⋅ Q π θ ( S t , A t ) | S 0 = s 0 ] {\displaystyle \nabla _{\theta }J(\theta )=E_{\pi _{\theta }}\left[\sum _{0\leq t\leq T}\gamma ^{t}\nabla _{\theta }\ln \pi _{\theta }(A_{t}|S_{t})\cdot Q^{\pi _{\theta }}(S_{t},A_{t}){\Big |}S_{0}=s_{0}\right]} by a similar argument using the tower law. Subtracting the value function as a baseline, we find that the advantage function A π ( S , A ) = Q π ( S , A ) − V π ( S ) {\displaystyle A^{\pi }(S,A)=Q^{\pi }(S,A)-V^{\pi }(S)} can be used as the critic as well: ∇ θ J ( θ ) = E π θ [ ∑ 0 ≤ t ≤ T γ t ∇ θ ln π θ ( A t | S t ) ⋅ A π θ ( S t , A t ) | S 0 = s 0 ] {\displaystyle \nabla _{\theta }J(\theta )=E_{\pi _{\theta }}\left[\sum _{0\leq t\leq T}\gamma ^{t}\nabla _{\theta }\ln \pi _{\theta }(A_{t}|S_{t})\cdot A^{\pi _{\theta }}(S_{t},A_{t}){\Big |}S_{0}=s_{0}\right]} In summary, there are many unbiased estimators for ∇ θ J θ {\textstyle \nabla _{\theta }J_{\theta }} , all in the form of: ∇ θ J ( θ ) = E π θ [ ∑ 0 ≤ t ≤ T ∇ θ ln π θ ( A t | S t ) ⋅ Ψ t | S 0 = s 0 ] {\displaystyle \nabla _{\theta }J(\theta )=E_{\pi _{\theta }}\left[\su
Vowpal Wabbit
Vowpal Wabbit (VW) is an open-source fast online interactive machine learning system library and program developed originally at Yahoo! Research, and currently at Microsoft Research. It was started and is led by John Langford. Vowpal Wabbit's interactive learning support is particularly notable including Contextual Bandits, Active Learning, and forms of guided Reinforcement Learning. Vowpal Wabbit provides an efficient scalable out-of-core implementation with support for a number of machine learning reductions, importance weighting, and a selection of different loss functions and optimization algorithms. == Notable features == The VW program supports: Multiple supervised (and semi-supervised) learning problems: Classification (both binary and multi-class) Regression Active learning (partially labeled data) for both regression and classification Multiple learning algorithms (model-types / representations) OLS regression Matrix factorization (sparse matrix SVD) Single layer neural net (with user specified hidden layer node count) Searn (Search and Learn) Latent Dirichlet Allocation (LDA) Stagewise polynomial approximation Recommend top-K out of N One-against-all (OAA) and cost-sensitive OAA reduction for multi-class Weighted all pairs Contextual-bandit (with multiple exploration/exploitation strategies) Multiple loss functions: squared error quantile hinge logistic poisson Multiple optimization algorithms Stochastic gradient descent (SGD) BFGS Conjugate gradient Regularization (L1 norm, L2 norm, & elastic net regularization) Flexible input - input features may be: Binary Numerical Categorical (via flexible feature-naming and the hash trick) Can deal with missing values/sparse-features Other features On the fly generation of feature interactions (quadratic and cubic) On the fly generation of N-grams with optional skips (useful for word/language data-sets) Automatic test-set holdout and early termination on multiple passes bootstrapping User settable online learning progress report + auditing of the model Hyperparameter optimization == Scalability == Vowpal wabbit has been used to learn a tera-feature (1012) data-set on 1000 nodes in one hour. Its scalability is aided by several factors: Out-of-core online learning: no need to load all data into memory The hashing trick: feature identities are converted to a weight index via a hash (uses 32-bit MurmurHash3) Exploiting multi-core CPUs: parsing of input and learning are done in separate threads. Compiled C++ code
Construction robots
Construction robots are a subset of industrial robots used for building and infrastructure construction on site, or in the production of materials and components offsite. A 2021 survey said 55% of construction companies in the United States, Europe, and China used robots in some form. This figure, however, reflects reported use across the construction value chain rather than widespread deployment of robots on active construction sites. Real-world adoption remains limited, with many robotic systems confined to pilot projects, controlled environments, or specific task applications rather than continuous on-site construction use. One of the main challenges in deploying robots on construction sites is the unstructured and variable nature of the environment, which differs fundamentally from controlled factory settings where industrial robots have traditionally operated. Some robots currently deployed on job sites assist with physically demanding or repetitive tasks: excavating, lifting heavy materials, surveying, laying out markers, tying rebar, and installing drywall. More advanced systems are being developed for exterior finishing, steel placement, masonry, and reinforced concrete work. In practice, rather than autonomous systems performing core building tasks, the most widely adopted robot applications on construction sites involve technologies such as aerial drones (or, less frequently, robot 'dogs' - for example, Boston Dynamics' Spot - or humanoid robots) used for surveying, inspection, and progress monitoring (the robots typically carry video and/or 360-degree cameras, LiDar scanners or other data capture devices, with data analysed using artificial intelligence and machine learning). Some emerging systems are designed as multifunctional construction robots, integrating multiple tools and capabilities within a single robotic platform to perform different stages of the construction process. These systems aim to improve operational flexibility and increase automation in complex construction environments. Experimental projects using robotic construction technologies and additive manufacturing have been demonstrated in several countries as part of broader efforts to industrialize the construction sector and improve productivity through automation and digitalization. == Features == Construction robots are generally required to meet the following criteria: Mobility: the ability to navigate around a construction site, including uneven terrain and confined spaces. Adaptability: the ability to handle components of variable size, weight, and shape. Environmental awareness: the ability to sense and respond to changing on-site conditions. Interactivity: the ability to operate alongside human workers and other equipment. Multitasking: the ability to perform several different operations within a single deployment. == Capabilities == Construction robots have been developed and tested for a range of on-site tasks, including: Progress monitoring — robots equipped with cameras and sensors can track construction progress and identify deviations from plans. Inspection — robots are used to investigate infrastructure at dangerous or inaccessible locations, reducing risk to human workers and eliminating human error. Wall construction — robotic systems can lay bricks and blocks with greater speed and consistency than manual labour. Earthmoving and material handling — autonomous excavators and haul trucks use GPS, lidar, and motion sensors to perform digging, trenching, and loading tasks with minimal human input. Grading and dozing — autonomous bulldozers use GPS, gyroscopes, and laser sensors to control blade angle and depth, improving surface finish accuracy and reducing material overuse. 3D printing — additive manufacturing systems can construct walls and structural elements directly from digital models. == Notable construction-related activities undertaken by robots == The distribution of robotic applications in construction varies across the project lifecycle. Most applications are concentrated in structural construction tasks such as masonry, concrete work, and assembly, while other phases, including planning, maintenance, and demolition, remain less represented. === Automated building systems === The Nisseki Yokohama Building (also known as Rail City Yokohama), a 30-storey office building in Yokohama, Japan, was constructed between 1994 and 1997 using the SMART system (Shimizu Manufacturing system by Advanced Robotics Technology), developed by Shimizu Corporation and a consortium of seven other Japanese companies. The system used automated horizontal hoists and vertical lifts to position steel beams, columns, precast concrete floor slabs, and prefabricated facade panels, with welding robots connecting structural elements under laser-guided precision. Each component was tracked by barcode to monitor progress and coordinate just-in-time delivery of materials. Obayashi Corporation developed the Advanced Building Construction System (ABCS), a similar automated platform used in several high-rise projects in Japan in the 1990s, including the NEC Head Office in Kanagawa (1997–2000). === Progress monitoring, inspection === Boston Dynamics' Spot was used in February 2024 to inspect sections of the M5 motorway in England for National Highways. A £15,000 humanoid robot (a G1 model from Chinese manufacturer Unitree) was deployed to capture 360-degree imagery and progress reports to support health and safety monitoring and reporting for UK contractor Tilbury Douglas in April 2026. In the US, Virginia Tech's ARCADE research lab is developing MARIO (Multi-Agent Robotic system for Inspection On-site), a heterogenous robotic system deploying multiple robots capable of different locomotion to perform remote real-time construction progress monitoring in complex construction sites. === Earthmoving === === Concrete works === Obayashi Corporation developed and deployed a robotic system for placing concrete layers in dam construction in Japan. A concrete floor finishing robot was deployed by Kajima and Tokimec in Japan. The MARK series were designed in 1984 to automate the levelling and trowelling of concrete slabs on construction sites, providing consistent finishing accuracy, improved efficiency, and reduced dependence on skilled labour === Masonry === SAM100 (Semi-Automated Mason), developed by Construction Robotics, is one of the first commercially available bricklaying robots for on-site masonry construction. In 2018, it was used in the construction of the University Arts Building at the University of Nevada, Reno — a $35.5 million facility — where it laid over 60,000 of the 100,000 bricks required, reducing the brick veneer installation time by approximately 50%. Hadrian X, developed by the Australian company Fastbrick Robotics, is a fully autonomous mobile bricklaying robot. In November 2022, it completed its first commercial project — five four-bedroom houses in Wellard, Western Australia. In February 2025, PulteGroup, one of the largest homebuilders in the United States, piloted Hadrian X on a site in Florida, constructing an entire house in a single day. === 3D printing === In May 2025, a residential building in Arinaga, Gran Canaria, Spain, was completed using 3D printing construction technology, as part of broader efforts to demonstrate robotic and additive manufacturing methods in the housing sector. In 2026, a three-storey apartment block in France was constructed using concrete 3D printing technology, three months faster than conventional building methods. Finland's Hyperion Robotics has opened a UK factory and used 3D printing with concrete to produce foundations for pipelines and for electricity substation bases, reducing time-consuming and weather-dependent onsite construction processes. == Social impact == The adoption of construction robots varies significantly by region and is shaped by labour market conditions, cultural attitudes, and regulatory frameworks. In Japan, construction robots have been embraced as a response to an ageing workforce and chronic labour shortages, and are generally viewed positively by the industry. In the United States, adoption has historically been slower, partly due to resistance from labour unions concerned about job displacement. Research suggests that the impact of automation on workers is uneven: while robots can create a productivity effect that benefits some workers, displacement effects are most pronounced among younger, less-educated workers in manufacturing-heavy regions. More than 60% of construction firms now report difficulty finding skilled operators, which has increased openness to automation as a practical solution to workforce shortages rather than a replacement for workers. In the UK, during onsite deployment of a humanoid robot for monitoring purposes, there were concerns that staff might think they were being watched ("It's not there to spy on people.... So, we insist that everyone is blurred out. N
ARKA descriptors in QSAR
In computational chemistry and cheminformatics, ARKA descriptors in QSAR are a class of molecular descriptors used in quantitative structure–activity relationship (QSAR) modeling (or related approaches such as QSPR and QSTR), a computational method for predicting the biological activity or toxicity of chemical compounds based on their molecular structure. Molecular descriptors are numerical values that summarize information about a molecule's structure, topology, geometry, or physicochemical properties in a form suitable for machine learning or statistical modeling. ARKA (Arithmetic Residuals in K-Groups Analysis) descriptors differ from traditional descriptors by encoding atomic-level information through recursive autoregression techniques, which aim to capture subtle structural patterns and improve predictive accuracy. They are designed to be both interpretable and well-suited to modeling nonlinear relationships in QSAR studies. == Comparisons == While QSAR is essentially a similarity-based approach, the occurrence of activity/property cliffs may greatly reduce the predictive accuracy of the developed models. The novel Arithmetic Residuals in K-groups Analysis (ARKA) approach is a supervised dimensionality reduction technique developed by the DTC Laboratory, Jadavpur University that can easily identify activity cliffs in a data set. Activity cliffs are similar in their structures but differ considerably in their activity. The basic idea of the ARKA descriptors is to group the conventional QSAR descriptors based on a predefined criterion and then assign weightage to each descriptor in each group. ARKA descriptors have also been used to develop classification-based and regression-based QSAR models with acceptable quality statistics. The ARKA descriptors have been used for the identification of activity cliffs in QSAR studies and/or model development by multiple researchers. A tutorial presentation on the ARKA descriptors is available. Recently a multi-class ARKA framework has been proposed for improved q-RASAR model generation.